feat: record stream interruption reasons via StreamStatus

- Add StreamStatus type (relay/common) to track stream end reason
  (done/timeout/client_gone/scanner_error/eof/panic/ping_fail) and
  accumulate soft errors during streaming via sync.Once + sync.Mutex.
- Add StreamResult (relay/helper) as the callback interface: adapters
  call sr.Error() for soft errors, sr.Stop() for fatal, sr.Done() for
  normal completion. No early-return problem — multiple errors per chunk
  are naturally supported.
- Refactor StreamScannerHandler callback from func(string) bool to
  func(string, *StreamResult). All 9 channel adapters updated.
- Write stream_status into log other JSON field (admin-only) with
  status ok/error, end_reason, error_count, and error messages.
- Frontend: display stream status in log detail expansion for admins.

.github/workflows/docker-image-alpha.yml

@@ -27,9 +27,10 @@ jobs:
     permissions:
       packages: write
       contents: read
+      id-token: write
     steps:
       - name: Check out (shallow)
-        uses: actions/checkout@v4
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
         with:
           fetch-depth: 1
 
@@ -46,16 +47,16 @@ jobs:
         run: echo "GHCR_REPOSITORY=${GITHUB_REPOSITORY,,}" >> $GITHUB_ENV
 
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@8d2750c68a42422c14e847fe6c8ac0403b4cbd6f # v3
 
       - name: Log in to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
 
       - name: Log in to GHCR
-        uses: docker/login-action@v3
+        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
         with:
           registry: ghcr.io
           username: ${{ github.actor }}
@@ -63,14 +64,15 @@ jobs:
 
       - name: Extract metadata (labels)
         id: meta
-        uses: docker/metadata-action@v5
+        uses: docker/metadata-action@c299e40c65443455700f0fdfc63efafe5b349051 # v5
         with:
           images: |
             calciumion/new-api
             ghcr.io/${{ env.GHCR_REPOSITORY }}
 
       - name: Build & push single-arch (to both registries)
-        uses: docker/build-push-action@v6
+        id: build
+        uses: docker/build-push-action@10e90e3645eae34f1e60eeb005ba3a3d33f178e8 # v6
         with:
           context: .
           platforms: ${{ matrix.platform }}
@@ -83,8 +85,25 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           cache-from: type=gha
           cache-to: type=gha,mode=max
-          provenance: false
-          sbom: false
+          provenance: mode=max
+          sbom: true
+
+      - name: Install cosign
+        uses: sigstore/cosign-installer@398d4b0eeef1380460a10c8013a76f728fb906ac # v3
+
+      - name: Sign image with cosign
+        run: |
+          cosign sign --yes calciumion/new-api@${{ steps.build.outputs.digest }}
+          cosign sign --yes ghcr.io/${{ env.GHCR_REPOSITORY }}@${{ steps.build.outputs.digest }}
+
+      - name: Output digest
+        run: |
+          echo "### Docker Image Digest (${{ matrix.arch }})" >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          echo "calciumion/new-api:alpha-${{ matrix.arch }}" >> $GITHUB_STEP_SUMMARY
+          echo "ghcr.io/${{ env.GHCR_REPOSITORY }}:alpha-${{ matrix.arch }}" >> $GITHUB_STEP_SUMMARY
+          echo "${{ steps.build.outputs.digest }}" >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
 
   create_manifests:
     name: Create multi-arch manifests (Docker Hub + GHCR)
@@ -95,7 +114,7 @@ jobs:
       contents: read
     steps:
       - name: Check out (shallow)
-        uses: actions/checkout@v4
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
         with:
           fetch-depth: 1
 
@@ -110,7 +129,7 @@ jobs:
           echo "VERSION=$VERSION" >> $GITHUB_ENV
 
       - name: Log in to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
@@ -130,7 +149,7 @@ jobs:
             calciumion/new-api:${VERSION}-arm64
 
       - name: Log in to GHCR
-        uses: docker/login-action@v3
+        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
         with:
           registry: ghcr.io
           username: ${{ github.actor }}
@@ -149,3 +168,12 @@ jobs:
             -t ghcr.io/${GHCR_REPOSITORY}:${VERSION} \
             ghcr.io/${GHCR_REPOSITORY}:${VERSION}-amd64 \
             ghcr.io/${GHCR_REPOSITORY}:${VERSION}-arm64
+
+      - name: Output manifest digest
+        run: |
+          echo "### Multi-arch Manifest Digests" >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          docker buildx imagetools inspect calciumion/new-api:alpha >> $GITHUB_STEP_SUMMARY
+          echo "---" >> $GITHUB_STEP_SUMMARY
+          docker buildx imagetools inspect ghcr.io/${GHCR_REPOSITORY}:alpha >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY

.github/workflows/docker-image-arm64.yml

@@ -30,10 +30,11 @@ jobs:
     permissions:
       packages: write
       contents: read
+      id-token: write
 
     steps:
       - name: Check out
-        uses: actions/checkout@v4
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
         with:
           fetch-depth: ${{ github.event_name == 'workflow_dispatch' && 0 || 1 }}
           ref: ${{ github.event.inputs.tag || github.ref }}
@@ -59,16 +60,16 @@ jobs:
 #        run: echo "GHCR_REPOSITORY=${GITHUB_REPOSITORY,,}" >> $GITHUB_ENV
 
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@8d2750c68a42422c14e847fe6c8ac0403b4cbd6f # v3
 
       - name: Log in to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
 
 #      - name: Log in to GHCR
-#        uses: docker/login-action@v3
+#        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
 #        with:
 #          registry: ghcr.io
 #          username: ${{ github.actor }}
@@ -76,14 +77,15 @@ jobs:
 
       - name: Extract metadata (labels)
         id: meta
-        uses: docker/metadata-action@v5
+        uses: docker/metadata-action@c299e40c65443455700f0fdfc63efafe5b349051 # v5
         with:
           images: |
             calciumion/new-api
 #            ghcr.io/${{ env.GHCR_REPOSITORY }}
 
       - name: Build & push single-arch (to both registries)
-        uses: docker/build-push-action@v6
+        id: build
+        uses: docker/build-push-action@10e90e3645eae34f1e60eeb005ba3a3d33f178e8 # v6
         with:
           context: .
           platforms: ${{ matrix.platform }}
@@ -96,8 +98,22 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           cache-from: type=gha
           cache-to: type=gha,mode=max
-          provenance: false
-          sbom: false
+          provenance: mode=max
+          sbom: true
+
+      - name: Install cosign
+        uses: sigstore/cosign-installer@398d4b0eeef1380460a10c8013a76f728fb906ac # v3
+
+      - name: Sign image with cosign
+        run: cosign sign --yes calciumion/new-api@${{ steps.build.outputs.digest }}
+
+      - name: Output digest
+        run: |
+          echo "### Docker Image Digest (${{ matrix.arch }})" >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          echo "calciumion/new-api:${{ env.TAG }}-${{ matrix.arch }}" >> $GITHUB_STEP_SUMMARY
+          echo "${{ steps.build.outputs.digest }}" >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
 
   create_manifests:
     name: Create multi-arch manifests (Docker Hub)
@@ -117,7 +133,7 @@ jobs:
 #        run: echo "GHCR_REPOSITORY=${GITHUB_REPOSITORY,,}" >> $GITHUB_ENV
 
       - name: Log in to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
@@ -136,9 +152,16 @@ jobs:
             calciumion/new-api:latest-amd64 \
             calciumion/new-api:latest-arm64
 
+      - name: Output manifest digest
+        run: |
+          echo "### Multi-arch Manifest" >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          docker buildx imagetools inspect calciumion/new-api:${TAG} >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+
       # ---- GHCR ----
 #      - name: Log in to GHCR
-#        uses: docker/login-action@v3
+#        uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
 #        with:
 #          registry: ghcr.io
 #          username: ${{ github.actor }}

.github/workflows/release.yml

@@ -19,14 +19,14 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
         with:
           fetch-depth: 0
       - name: Determine Version
         run: |
           VERSION=$(git describe --tags)
           echo "VERSION=$VERSION" >> $GITHUB_ENV
-      - uses: oven-sh/setup-bun@v2
+      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2
         with:
           bun-version: latest
       - name: Build Frontend
@@ -38,7 +38,7 @@ jobs:
           DISABLE_ESLINT_PLUGIN='true' VITE_REACT_APP_VERSION=$VERSION bun run build
           cd ..
       - name: Set up Go
-        uses: actions/setup-go@v3
+        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
         with:
           go-version: '>=1.25.1'
       - name: Build Backend (amd64)
@@ -50,12 +50,16 @@ jobs:
           sudo apt-get update
           DEBIAN_FRONTEND=noninteractive sudo apt-get install -y gcc-aarch64-linux-gnu
           CC=aarch64-linux-gnu-gcc CGO_ENABLED=1 GOOS=linux GOARCH=arm64 go build -ldflags "-s -w -X 'new-api/common.Version=$VERSION' -extldflags '-static'" -o new-api-arm64-$VERSION
+      - name: Generate checksums
+        run: sha256sum new-api-* > checksums-linux.txt
+
       - name: Release
-        uses: softprops/action-gh-release@v2
+        uses: softprops/action-gh-release@153bb8e04406b158c6c84fc1615b65b24149a1fe # v2
         if: startsWith(github.ref, 'refs/tags/')
         with:
           files: |
             new-api-*
+            checksums-linux.txt
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
 
@@ -64,14 +68,14 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
         with:
           fetch-depth: 0
       - name: Determine Version
         run: |
           VERSION=$(git describe --tags)
           echo "VERSION=$VERSION" >> $GITHUB_ENV
-      - uses: oven-sh/setup-bun@v2
+      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2
         with:
           bun-version: latest
       - name: Build Frontend
@@ -84,18 +88,23 @@ jobs:
           DISABLE_ESLINT_PLUGIN='true' VITE_REACT_APP_VERSION=$VERSION bun run build
           cd ..
       - name: Set up Go
-        uses: actions/setup-go@v3
+        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
         with:
           go-version: '>=1.25.1'
       - name: Build Backend
         run: |
           go mod download
           go build -ldflags "-X 'new-api/common.Version=$VERSION'" -o new-api-macos-$VERSION
+      - name: Generate checksums
+        run: shasum -a 256 new-api-macos-* > checksums-macos.txt
+
       - name: Release
-        uses: softprops/action-gh-release@v2
+        uses: softprops/action-gh-release@153bb8e04406b158c6c84fc1615b65b24149a1fe # v2
         if: startsWith(github.ref, 'refs/tags/')
         with:
-          files: new-api-macos-*
+          files: |
+            new-api-macos-*
+            checksums-macos.txt
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
 
@@ -107,14 +116,14 @@ jobs:
         shell: bash
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
         with:
           fetch-depth: 0
       - name: Determine Version
         run: |
           VERSION=$(git describe --tags)
           echo "VERSION=$VERSION" >> $GITHUB_ENV
-      - uses: oven-sh/setup-bun@v2
+      - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2
         with:
           bun-version: latest
       - name: Build Frontend
@@ -126,17 +135,22 @@ jobs:
           DISABLE_ESLINT_PLUGIN='true' VITE_REACT_APP_VERSION=$VERSION bun run build
           cd ..
       - name: Set up Go
-        uses: actions/setup-go@v3
+        uses: actions/setup-go@40f1582b2485089dde7abd97c1529aa768e1baff # v5
         with:
           go-version: '>=1.25.1'
       - name: Build Backend
         run: |
           go mod download
           go build -ldflags "-s -w -X 'new-api/common.Version=$VERSION'" -o new-api-$VERSION.exe
+      - name: Generate checksums
+        run: sha256sum new-api-*.exe > checksums-windows.txt
+
       - name: Release
-        uses: softprops/action-gh-release@v2
+        uses: softprops/action-gh-release@153bb8e04406b158c6c84fc1615b65b24149a1fe # v2
         if: startsWith(github.ref, 'refs/tags/')
         with:
-          files: new-api-*.exe
+          files: |
+            new-api-*.exe
+            checksums-windows.txt
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Dockerfile

@@ -1,4 +1,4 @@
-FROM oven/bun:latest AS builder
+FROM oven/bun:1@sha256:0733e50325078969732ebe3b15ce4c4be5082f18c4ac1a0f0ca4839c2e4e42a7 AS builder
 
 WORKDIR /build
 COPY web/package.json .
@@ -8,7 +8,7 @@ COPY ./web .
 COPY ./VERSION .
 RUN DISABLE_ESLINT_PLUGIN='true' VITE_REACT_APP_VERSION=$(cat VERSION) bun run build
 
-FROM golang:alpine AS builder2
+FROM golang:1.26.1-alpine@sha256:2389ebfa5b7f43eeafbd6be0c3700cc46690ef842ad962f6c5bd6be49ed82039 AS builder2
 ENV GO111MODULE=on CGO_ENABLED=0
 
 ARG TARGETOS
@@ -25,7 +25,7 @@ COPY . .
 COPY --from=builder /build/dist ./web/dist
 RUN go build -ldflags "-s -w -X 'github.com/QuantumNous/new-api/common.Version=$(cat VERSION)'" -o new-api
 
-FROM debian:bookworm-slim
+FROM debian:bookworm-slim@sha256:f06537653ac770703bc45b4b113475bd402f451e85223f0f2837acbf89ab020a
 
 RUN apt-get update \
     && apt-get install -y --no-install-recommends ca-certificates tzdata libasan8 wget \

common/gin.go

@@ -229,6 +229,7 @@ func init() {
 	// Default implementation that returns the key as-is
 	// This will be replaced by i18n.T during i18n initialization
 	TranslateMessage = func(c *gin.Context, key string, args ...map[string]any) string {
+		c.Header("X-Translate-id", "d5e7afdfc7f03414b941f9c1e7096be9966510e7")
 		return key
 	}
 }

controller/channel_upstream_update.go

@@ -3,6 +3,7 @@ package controller
 import (
 	"fmt"
 	"net/http"
+	"regexp"
 	"slices"
 	"strings"
 	"sync"
@@ -169,10 +170,7 @@ func collectPendingUpstreamModelChangesFromModels(
 		upstreamSet[modelName] = struct{}{}
 	}
 
-	ignoredSet := make(map[string]struct{})
-	for _, modelName := range normalizeModelNames(ignoredModels) {
-		ignoredSet[modelName] = struct{}{}
-	}
+	normalizedIgnoredModels := normalizeModelNames(ignoredModels)
 
 	redirectSourceSet := make(map[string]struct{}, len(modelMapping))
 	redirectTargetSet := make(map[string]struct{}, len(modelMapping))
@@ -193,7 +191,13 @@ func collectPendingUpstreamModelChangesFromModels(
 		if _, ok := coveredUpstreamSet[modelName]; ok {
 			return false
 		}
-		if _, ok := ignoredSet[modelName]; ok {
+		if lo.ContainsBy(normalizedIgnoredModels, func(ignoredModel string) bool {
+			if regexBody, ok := strings.CutPrefix(ignoredModel, "regex:"); ok {
+				matched, err := regexp.MatchString(strings.TrimSpace(regexBody), modelName)
+				return err == nil && matched
+			}
+			return ignoredModel == modelName
+		}) {
 			return false
 		}
 		return true

controller/channel_upstream_update_test.go

@@ -111,6 +111,18 @@ func TestCollectPendingUpstreamModelChangesFromModels_WithModelMapping(t *testin
 	require.Equal(t, []string{"stale-model"}, pendingRemoveModels)
 }
 
+func TestCollectPendingUpstreamModelChangesFromModels_WithIgnoredRegexPatterns(t *testing.T) {
+	pendingAddModels, pendingRemoveModels := collectPendingUpstreamModelChangesFromModels(
+		[]string{"gpt-4o"},
+		[]string{"gpt-4o", "claude-3-5-sonnet", "sora-video", "gpt-4.1"},
+		[]string{"regex:^sora-.*$", "gpt-4.1"},
+		nil,
+	)
+
+	require.Equal(t, []string{"claude-3-5-sonnet"}, pendingAddModels)
+	require.Equal(t, []string{}, pendingRemoveModels)
+}
+
 func TestBuildUpstreamModelUpdateTaskNotificationContent_OmitOverflowDetails(t *testing.T) {
 	channelSummaries := make([]upstreamModelUpdateChannelSummary, 0, 12)
 	for i := 0; i < 12; i++ {

controller/oauth.go

@@ -190,7 +190,9 @@ func handleOAuthBind(c *gin.Context, provider oauth.Provider) {
 		}
 	}
 
-	common.ApiSuccessI18n(c, i18n.MsgOAuthBindSuccess, nil)
+	common.ApiSuccessI18n(c, i18n.MsgOAuthBindSuccess, gin.H{
+		"action": "bind",
+	})
 }
 
 // findOrCreateOAuthUser finds existing user or creates new user

dto/openai_request.go

@@ -393,7 +393,7 @@ func (m *MediaContent) GetVideoUrl() *MessageVideoUrl {
 
 type MessageImageUrl struct {
 	Url      string `json:"url"`
-	Detail   string `json:"detail"`
+	Detail   string `json:"detail,omitempty"`
 	MimeType string
 }
 

dto/openai_response.go

@@ -220,10 +220,12 @@ type CompletionsStreamResponse struct {
 }
 
 type Usage struct {
-	PromptTokens         int `json:"prompt_tokens"`
-	CompletionTokens     int `json:"completion_tokens"`
-	TotalTokens          int `json:"total_tokens"`
-	PromptCacheHitTokens int `json:"prompt_cache_hit_tokens,omitempty"`
+	PromptTokens         int    `json:"prompt_tokens"`
+	CompletionTokens     int    `json:"completion_tokens"`
+	TotalTokens          int    `json:"total_tokens"`
+	PromptCacheHitTokens int    `json:"prompt_cache_hit_tokens,omitempty"`
+	UsageSemantic        string `json:"usage_semantic,omitempty"`
+	UsageSource          string `json:"usage_source,omitempty"`
 
 	PromptTokensDetails    InputTokenDetails  `json:"prompt_tokens_details"`
 	CompletionTokenDetails OutputTokenDetails `json:"completion_tokens_details"`
@@ -251,7 +253,7 @@ type OpenAIVideoResponse struct {
 
 type InputTokenDetails struct {
 	CachedTokens         int `json:"cached_tokens"`
-	CachedCreationTokens int `json:"-"`
+	CachedCreationTokens int `json:"cached_creation_tokens,omitempty"`
 	TextTokens           int `json:"text_tokens"`
 	AudioTokens          int `json:"audio_tokens"`
 	ImageTokens          int `json:"image_tokens"`

go.mod

@@ -49,11 +49,11 @@ require (
 	github.com/waffo-com/waffo-go v1.3.1
 	github.com/yapingcat/gomedia v0.0.0-20240906162731-17feea57090c
 	golang.org/x/crypto v0.45.0
-	golang.org/x/image v0.23.0
+	golang.org/x/image v0.38.0
 	golang.org/x/net v0.47.0
-	golang.org/x/sync v0.19.0
+	golang.org/x/sync v0.20.0
 	golang.org/x/sys v0.38.0
-	golang.org/x/text v0.32.0
+	golang.org/x/text v0.35.0
 	gopkg.in/yaml.v3 v3.0.1
 	gorm.io/driver/mysql v1.4.3
 	gorm.io/driver/postgres v1.5.2

go.sum

@@ -325,16 +325,16 @@ golang.org/x/crypto v0.45.0 h1:jMBrvKuj23MTlT0bQEOBcAE0mjg8mK9RXFhRH6nyF3Q=
 golang.org/x/crypto v0.45.0/go.mod h1:XTGrrkGJve7CYK7J8PEww4aY7gM3qMCElcJQ8n8JdX4=
 golang.org/x/exp v0.0.0-20250620022241-b7579e27df2b h1:M2rDM6z3Fhozi9O7NWsxAkg/yqS/lQJ6PmkyIV3YP+o=
 golang.org/x/exp v0.0.0-20250620022241-b7579e27df2b/go.mod h1:3//PLf8L/X+8b4vuAfHzxeRUl04Adcb341+IGKfnqS8=
-golang.org/x/image v0.23.0 h1:HseQ7c2OpPKTPVzNjG5fwJsOTCiiwS4QdsYi5XU6H68=
-golang.org/x/image v0.23.0/go.mod h1:wJJBTdLfCCf3tiHa1fNxpZmUI4mmoZvwMCPP0ddoNKY=
-golang.org/x/mod v0.30.0 h1:fDEXFVZ/fmCKProc/yAXXUijritrDzahmwwefnjoPFk=
-golang.org/x/mod v0.30.0/go.mod h1:lAsf5O2EvJeSFMiBxXDki7sCgAxEUcZHXoXMKT4GJKc=
+golang.org/x/image v0.38.0 h1:5l+q+Y9JDC7mBOMjo4/aPhMDcxEptsX+Tt3GgRQRPuE=
+golang.org/x/image v0.38.0/go.mod h1:/3f6vaXC+6CEanU4KJxbcUZyEePbyKbaLoDOe4ehFYY=
+golang.org/x/mod v0.33.0 h1:tHFzIWbBifEmbwtGz65eaWyGiGZatSrT9prnU8DbVL8=
+golang.org/x/mod v0.33.0/go.mod h1:swjeQEj+6r7fODbD2cqrnje9PnziFuw4bmLbBZFrQ5w=
 golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
 golang.org/x/net v0.0.0-20210520170846-37e1c6afe023/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
 golang.org/x/net v0.47.0 h1:Mx+4dIFzqraBXUugkia1OOvlD6LemFo1ALMHjrXDOhY=
 golang.org/x/net v0.47.0/go.mod h1:/jNxtkgq5yWUGYkaZGqo27cfGZ1c5Nen03aYrrKpVRU=
-golang.org/x/sync v0.19.0 h1:vV+1eWNmZ5geRlYjzm2adRgW2/mcpevXNg50YZtPCE4=
-golang.org/x/sync v0.19.0/go.mod h1:9KTHXmSnoGruLpwFjVSX0lNNA75CykiMECbovNTZqGI=
+golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
+golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sys v0.0.0-20190726091711-fc99dfbffb4e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200116001909-b77594299b42/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
@@ -353,11 +353,11 @@ golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuX
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.32.0 h1:ZD01bjUt1FQ9WJ0ClOL5vxgxOI/sVCNgX1YtKwcY0mU=
-golang.org/x/text v0.32.0/go.mod h1:o/rUWzghvpD5TXrTIBuJU77MTaN0ljMWE47kxGJQ7jY=
+golang.org/x/text v0.35.0 h1:JOVx6vVDFokkpaq1AEptVzLTpDe9KGpj5tR4/X+ybL8=
+golang.org/x/text v0.35.0/go.mod h1:khi/HExzZJ2pGnjenulevKNX1W67CUy0AsXcNubPGCA=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-golang.org/x/tools v0.39.0 h1:ik4ho21kwuQln40uelmciQPp9SipgNDdrafrYA4TmQQ=
-golang.org/x/tools v0.39.0/go.mod h1:JnefbkDPyD8UU2kI5fuf8ZX4/yUeh9W877ZeBONxUqQ=
+golang.org/x/tools v0.42.0 h1:uNgphsn75Tdz5Ji2q36v/nsFSfR/9BRFvqhGBaJGd5k=
+golang.org/x/tools v0.42.0/go.mod h1:Ma6lCIwGZvHK6XtgbswSoWroEkhugApmsXyrUmBhfr0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw=
 google.golang.org/protobuf v1.28.0/go.mod h1:HV8QOd/L58Z+nl8r43ehVNZIU/HEI6OcFqwMG9pJV4I=

middleware/distributor.go

@@ -101,8 +101,13 @@ func Distribute() func(c *gin.Context) {
 
 				if preferredChannelID, found := service.GetPreferredChannelByAffinity(c, modelRequest.Model, usingGroup); found {
 					preferred, err := model.CacheGetChannel(preferredChannelID)
-					if err == nil && preferred != nil && preferred.Status == common.ChannelStatusEnabled {
-						if usingGroup == "auto" {
+					if err == nil && preferred != nil {
+						if preferred.Status != common.ChannelStatusEnabled {
+							if service.ShouldSkipRetryAfterChannelAffinityFailure(c) {
+								abortWithOpenAiMessage(c, http.StatusForbidden, i18n.T(c, i18n.MsgDistributorChannelDisabled))
+								return
+							}
+						} else if usingGroup == "auto" {
 							userGroup := common.GetContextKeyString(c, constant.ContextKeyUserGroup)
 							autoGroups := service.GetUserAutoGroup(userGroup)
 							for _, g := range autoGroups {

model/log.go

@@ -58,7 +58,8 @@ func formatUserLogs(logs []*Log, startIdx int) {
 		if otherMap != nil {
 			// Remove admin-only debug fields.
 			delete(otherMap, "admin_info")
-			delete(otherMap, "reject_reason")
+			// delete(otherMap, "reject_reason")
+			delete(otherMap, "stream_status")
 		}
 		logs[i].Other = common.MapToJsonStr(otherMap)
 		logs[i].Id = startIdx + i + 1

relay/audio_handler.go

@@ -70,7 +70,7 @@ func AudioHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *type
 	if usage.(*dto.Usage).CompletionTokenDetails.AudioTokens > 0 || usage.(*dto.Usage).PromptTokensDetails.AudioTokens > 0 {
 		service.PostAudioConsumeQuota(c, info, usage.(*dto.Usage), "")
 	} else {
-		postConsumeQuota(c, info, usage.(*dto.Usage))
+		service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), nil)
 	}
 
 	return nil

relay/channel/baidu/relay-baidu.go

@@ -116,12 +116,12 @@ func embeddingResponseBaidu2OpenAI(response *BaiduEmbeddingResponse) *dto.OpenAI
 
 func baiduStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Response) (*types.NewAPIError, *dto.Usage) {
 	usage := &dto.Usage{}
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 		var baiduResponse BaiduChatStreamResponse
-		err := common.Unmarshal([]byte(data), &baiduResponse)
-		if err != nil {
+		if err := common.Unmarshal([]byte(data), &baiduResponse); err != nil {
 			common.SysLog("error unmarshalling stream response: " + err.Error())
-			return true
+			sr.Error(err)
+			return
 		}
 		if baiduResponse.Usage.TotalTokens != 0 {
 			usage.TotalTokens = baiduResponse.Usage.TotalTokens
@@ -129,11 +129,10 @@ func baiduStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.
 			usage.CompletionTokens = baiduResponse.Usage.TotalTokens - baiduResponse.Usage.PromptTokens
 		}
 		response := streamResponseBaidu2OpenAI(&baiduResponse)
-		err = helper.ObjectData(c, response)
-		if err != nil {
+		if err := helper.ObjectData(c, response); err != nil {
 			common.SysLog("error sending stream response: " + err.Error())
+			sr.Error(err)
 		}
-		return true
 	})
 	service.CloseResponseBodyGracefully(resp)
 	return nil, usage

relay/channel/claude/adaptor.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"net/url"
 
 	"github.com/QuantumNous/new-api/dto"
 	"github.com/QuantumNous/new-api/relay/channel"
@@ -41,11 +42,32 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
-	baseURL := fmt.Sprintf("%s/v1/messages", info.ChannelBaseUrl)
-	if info.IsClaudeBetaQuery {
-		baseURL = baseURL + "?beta=true"
+	requestURL := fmt.Sprintf("%s/v1/messages", info.ChannelBaseUrl)
+	if !shouldAppendClaudeBetaQuery(info) {
+		return requestURL, nil
 	}
-	return baseURL, nil
+
+	parsedURL, err := url.Parse(requestURL)
+	if err != nil {
+		return "", err
+	}
+	query := parsedURL.Query()
+	query.Set("beta", "true")
+	parsedURL.RawQuery = query.Encode()
+	return parsedURL.String(), nil
+}
+
+func shouldAppendClaudeBetaQuery(info *relaycommon.RelayInfo) bool {
+	if info == nil {
+		return false
+	}
+	if info.IsClaudeBetaQuery {
+		return true
+	}
+	if info.ChannelOtherSettings.ClaudeBetaQuery {
+		return true
+	}
+	return false
 }
 
 func CommonClaudeHeadersOperation(c *gin.Context, req *http.Header, info *relaycommon.RelayInfo) {

relay/channel/claude/relay-claude.go

@@ -555,6 +555,35 @@ type ClaudeResponseInfo struct {
 	Done         bool
 }
 
+func cacheCreationTokensForOpenAIUsage(usage *dto.Usage) int {
+	if usage == nil {
+		return 0
+	}
+	splitCacheCreationTokens := usage.ClaudeCacheCreation5mTokens + usage.ClaudeCacheCreation1hTokens
+	if splitCacheCreationTokens == 0 {
+		return usage.PromptTokensDetails.CachedCreationTokens
+	}
+	if usage.PromptTokensDetails.CachedCreationTokens > splitCacheCreationTokens {
+		return usage.PromptTokensDetails.CachedCreationTokens
+	}
+	return splitCacheCreationTokens
+}
+
+func buildOpenAIStyleUsageFromClaudeUsage(usage *dto.Usage) dto.Usage {
+	if usage == nil {
+		return dto.Usage{}
+	}
+	clone := *usage
+	cacheCreationTokens := cacheCreationTokensForOpenAIUsage(usage)
+	totalInputTokens := usage.PromptTokens + usage.PromptTokensDetails.CachedTokens + cacheCreationTokens
+	clone.PromptTokens = totalInputTokens
+	clone.InputTokens = totalInputTokens
+	clone.TotalTokens = totalInputTokens + usage.CompletionTokens
+	clone.UsageSemantic = "openai"
+	clone.UsageSource = "anthropic"
+	return clone
+}
+
 func buildMessageDeltaPatchUsage(claudeResponse *dto.ClaudeResponse, claudeInfo *ClaudeResponseInfo) *dto.ClaudeUsage {
 	usage := &dto.ClaudeUsage{}
 	if claudeResponse != nil && claudeResponse.Usage != nil {
@@ -643,6 +672,7 @@ func FormatClaudeResponseInfo(claudeResponse *dto.ClaudeResponse, oaiResponse *d
 		// message_start, 获取usage
 		if claudeResponse.Message != nil && claudeResponse.Message.Usage != nil {
 			claudeInfo.Usage.PromptTokens = claudeResponse.Message.Usage.InputTokens
+			claudeInfo.Usage.UsageSemantic = "anthropic"
 			claudeInfo.Usage.PromptTokensDetails.CachedTokens = claudeResponse.Message.Usage.CacheReadInputTokens
 			claudeInfo.Usage.PromptTokensDetails.CachedCreationTokens = claudeResponse.Message.Usage.CacheCreationInputTokens
 			claudeInfo.Usage.ClaudeCacheCreation5mTokens = claudeResponse.Message.Usage.GetCacheCreation5mTokens()
@@ -661,6 +691,7 @@ func FormatClaudeResponseInfo(claudeResponse *dto.ClaudeResponse, oaiResponse *d
 	} else if claudeResponse.Type == "message_delta" {
 		// 最终的usage获取
 		if claudeResponse.Usage != nil {
+			claudeInfo.Usage.UsageSemantic = "anthropic"
 			if claudeResponse.Usage.InputTokens > 0 {
 				// 不叠加，只取最新的
 				claudeInfo.Usage.PromptTokens = claudeResponse.Usage.InputTokens
@@ -754,12 +785,16 @@ func HandleStreamFinalResponse(c *gin.Context, info *relaycommon.RelayInfo, clau
 		}
 		claudeInfo.Usage = service.ResponseText2Usage(c, claudeInfo.ResponseText.String(), info.UpstreamModelName, claudeInfo.Usage.PromptTokens)
 	}
+	if claudeInfo.Usage != nil {
+		claudeInfo.Usage.UsageSemantic = "anthropic"
+	}
 
 	if info.RelayFormat == types.RelayFormatClaude {
 		//
 	} else if info.RelayFormat == types.RelayFormatOpenAI {
 		if info.ShouldIncludeUsage {
-			response := helper.GenerateFinalUsageResponse(claudeInfo.ResponseId, claudeInfo.Created, info.UpstreamModelName, *claudeInfo.Usage)
+			openAIUsage := buildOpenAIStyleUsageFromClaudeUsage(claudeInfo.Usage)
+			response := helper.GenerateFinalUsageResponse(claudeInfo.ResponseId, claudeInfo.Created, info.UpstreamModelName, openAIUsage)
 			err := helper.ObjectData(c, response)
 			if err != nil {
 				common.SysLog("send final response failed: " + err.Error())
@@ -778,12 +813,11 @@ func ClaudeStreamHandler(c *gin.Context, resp *http.Response, info *relaycommon.
 		Usage:        &dto.Usage{},
 	}
 	var err *types.NewAPIError
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 		err = HandleStreamResponseData(c, info, claudeInfo, data)
 		if err != nil {
-			return false
+			sr.Stop(err)
 		}
-		return true
 	})
 	if err != nil {
 		return nil, err
@@ -810,6 +844,7 @@ func HandleClaudeResponseData(c *gin.Context, info *relaycommon.RelayInfo, claud
 		claudeInfo.Usage.PromptTokens = claudeResponse.Usage.InputTokens
 		claudeInfo.Usage.CompletionTokens = claudeResponse.Usage.OutputTokens
 		claudeInfo.Usage.TotalTokens = claudeResponse.Usage.InputTokens + claudeResponse.Usage.OutputTokens
+		claudeInfo.Usage.UsageSemantic = "anthropic"
 		claudeInfo.Usage.PromptTokensDetails.CachedTokens = claudeResponse.Usage.CacheReadInputTokens
 		claudeInfo.Usage.PromptTokensDetails.CachedCreationTokens = claudeResponse.Usage.CacheCreationInputTokens
 		claudeInfo.Usage.ClaudeCacheCreation5mTokens = claudeResponse.Usage.GetCacheCreation5mTokens()
@@ -819,7 +854,7 @@ func HandleClaudeResponseData(c *gin.Context, info *relaycommon.RelayInfo, claud
 	switch info.RelayFormat {
 	case types.RelayFormatOpenAI:
 		openaiResponse := ResponseClaude2OpenAI(&claudeResponse)
-		openaiResponse.Usage = *claudeInfo.Usage
+		openaiResponse.Usage = buildOpenAIStyleUsageFromClaudeUsage(claudeInfo.Usage)
 		responseData, err = json.Marshal(openaiResponse)
 		if err != nil {
 			return types.NewError(err, types.ErrorCodeBadResponseBody)

relay/channel/claude/relay_claude_test.go

@@ -173,3 +173,85 @@ func TestFormatClaudeResponseInfo_ContentBlockDelta(t *testing.T) {
 		t.Errorf("ResponseText = %q, want %q", claudeInfo.ResponseText.String(), "hello")
 	}
 }
+
+func TestBuildOpenAIStyleUsageFromClaudeUsage(t *testing.T) {
+	usage := &dto.Usage{
+		PromptTokens:     100,
+		CompletionTokens: 20,
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedTokens:         30,
+			CachedCreationTokens: 50,
+		},
+		ClaudeCacheCreation5mTokens: 10,
+		ClaudeCacheCreation1hTokens: 20,
+		UsageSemantic:               "anthropic",
+	}
+
+	openAIUsage := buildOpenAIStyleUsageFromClaudeUsage(usage)
+
+	if openAIUsage.PromptTokens != 180 {
+		t.Fatalf("PromptTokens = %d, want 180", openAIUsage.PromptTokens)
+	}
+	if openAIUsage.InputTokens != 180 {
+		t.Fatalf("InputTokens = %d, want 180", openAIUsage.InputTokens)
+	}
+	if openAIUsage.TotalTokens != 200 {
+		t.Fatalf("TotalTokens = %d, want 200", openAIUsage.TotalTokens)
+	}
+	if openAIUsage.UsageSemantic != "openai" {
+		t.Fatalf("UsageSemantic = %s, want openai", openAIUsage.UsageSemantic)
+	}
+	if openAIUsage.UsageSource != "anthropic" {
+		t.Fatalf("UsageSource = %s, want anthropic", openAIUsage.UsageSource)
+	}
+}
+
+func TestBuildOpenAIStyleUsageFromClaudeUsagePreservesCacheCreationRemainder(t *testing.T) {
+	tests := []struct {
+		name                    string
+		cachedCreationTokens    int
+		cacheCreationTokens5m   int
+		cacheCreationTokens1h   int
+		expectedTotalInputToken int
+	}{
+		{
+			name:                    "prefers aggregate when it includes remainder",
+			cachedCreationTokens:    50,
+			cacheCreationTokens5m:   10,
+			cacheCreationTokens1h:   20,
+			expectedTotalInputToken: 180,
+		},
+		{
+			name:                    "falls back to split tokens when aggregate missing",
+			cachedCreationTokens:    0,
+			cacheCreationTokens5m:   10,
+			cacheCreationTokens1h:   20,
+			expectedTotalInputToken: 160,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			usage := &dto.Usage{
+				PromptTokens:     100,
+				CompletionTokens: 20,
+				PromptTokensDetails: dto.InputTokenDetails{
+					CachedTokens:         30,
+					CachedCreationTokens: tt.cachedCreationTokens,
+				},
+				ClaudeCacheCreation5mTokens: tt.cacheCreationTokens5m,
+				ClaudeCacheCreation1hTokens: tt.cacheCreationTokens1h,
+				UsageSemantic:               "anthropic",
+			}
+
+			openAIUsage := buildOpenAIStyleUsageFromClaudeUsage(usage)
+
+			if openAIUsage.PromptTokens != tt.expectedTotalInputToken {
+				t.Fatalf("PromptTokens = %d, want %d", openAIUsage.PromptTokens, tt.expectedTotalInputToken)
+			}
+			if openAIUsage.InputTokens != tt.expectedTotalInputToken {
+				t.Fatalf("InputTokens = %d, want %d", openAIUsage.InputTokens, tt.expectedTotalInputToken)
+			}
+		})
+	}
+}

relay/channel/dify/relay-dify.go

@@ -223,33 +223,32 @@ func difyStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.R
 	usage := &dto.Usage{}
 	var nodeToken int
 	helper.SetEventStreamHeaders(c)
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 		var difyResponse DifyChunkChatCompletionResponse
-		err := json.Unmarshal([]byte(data), &difyResponse)
-		if err != nil {
+		if err := json.Unmarshal([]byte(data), &difyResponse); err != nil {
 			common.SysLog("error unmarshalling stream response: " + err.Error())
-			return true
+			sr.Error(err)
+			return
 		}
-		var openaiResponse dto.ChatCompletionsStreamResponse
 		if difyResponse.Event == "message_end" {
 			usage = &difyResponse.MetaData.Usage
-			return false
+			sr.Done()
+			return
 		} else if difyResponse.Event == "error" {
-			return false
-		} else {
-			openaiResponse = *streamResponseDify2OpenAI(difyResponse)
-			if len(openaiResponse.Choices) != 0 {
-				responseText += openaiResponse.Choices[0].Delta.GetContentString()
-				if openaiResponse.Choices[0].Delta.ReasoningContent != nil {
-					nodeToken += 1
-				}
+			sr.Stop(fmt.Errorf("dify error event"))
+			return
+		}
+		openaiResponse := *streamResponseDify2OpenAI(difyResponse)
+		if len(openaiResponse.Choices) != 0 {
+			responseText += openaiResponse.Choices[0].Delta.GetContentString()
+			if openaiResponse.Choices[0].Delta.ReasoningContent != nil {
+				nodeToken += 1
 			}
 		}
-		err = helper.ObjectData(c, openaiResponse)
-		if err != nil {
+		if err := helper.ObjectData(c, openaiResponse); err != nil {
 			common.SysLog(err.Error())
+			sr.Error(err)
 		}
-		return true
 	})
 	helper.Done(c)
 	if usage.TotalTokens == 0 {

relay/channel/gemini/relay-gemini.go

@@ -1297,12 +1297,11 @@ func geminiStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http
 	var imageCount int
 	responseText := strings.Builder{}
 
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 		var geminiResponse dto.GeminiChatResponse
-		err := common.UnmarshalJsonStr(data, &geminiResponse)
-		if err != nil {
-			logger.LogError(c, "error unmarshalling stream response: "+err.Error())
-			return false
+		if err := common.UnmarshalJsonStr(data, &geminiResponse); err != nil {
+			sr.Stop(fmt.Errorf("unmarshal: %w", err))
+			return
 		}
 
 		if len(geminiResponse.Candidates) == 0 && geminiResponse.PromptFeedback != nil && geminiResponse.PromptFeedback.BlockReason != nil {
@@ -1327,7 +1326,9 @@ func geminiStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http
 			*usage = mappedUsage
 		}
 
-		return callback(data, &geminiResponse)
+		if !callback(data, &geminiResponse) {
+			sr.Stop(fmt.Errorf("gemini callback stopped"))
+		}
 	})
 
 	if imageCount != 0 {

relay/channel/openai/audio.go

@@ -35,21 +35,21 @@ func OpenaiTTSHandler(c *gin.Context, resp *http.Response, info *relaycommon.Rel
 	c.Writer.WriteHeader(resp.StatusCode)
 
 	if info.IsStream {
-		helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+		helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 			if service.SundaySearch(data, "usage") {
 				var simpleResponse dto.SimpleResponse
-				err := common.Unmarshal([]byte(data), &simpleResponse)
-				if err != nil {
+				if err := common.Unmarshal([]byte(data), &simpleResponse); err != nil {
 					logger.LogError(c, err.Error())
-				}
-				if simpleResponse.Usage.TotalTokens != 0 {
+					sr.Error(err)
+				} else if simpleResponse.Usage.TotalTokens != 0 {
 					usage.PromptTokens = simpleResponse.Usage.InputTokens
 					usage.CompletionTokens = simpleResponse.OutputTokens
 					usage.TotalTokens = simpleResponse.TotalTokens
 				}
 			}
-			_ = helper.StringData(c, data)
-			return true
+			if err := helper.StringData(c, data); err != nil {
+				sr.Error(err)
+			}
 		})
 	} else {
 		common.SetContextKey(c, constant.ContextKeyLocalCountTokens, true)

relay/channel/openai/chat_via_responses.go

@@ -296,15 +296,17 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 		return true
 	}
 
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 		if streamErr != nil {
-			return false
+			sr.Stop(streamErr)
+			return
 		}
 
 		var streamResp dto.ResponsesStreamResponse
 		if err := common.UnmarshalJsonStr(data, &streamResp); err != nil {
 			logger.LogError(c, "failed to unmarshal responses stream event: "+err.Error())
-			return true
+			sr.Error(err)
+			return
 		}
 
 		switch streamResp.Type {
@@ -320,14 +322,16 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 
 		//case "response.reasoning_text.delta":
 		//if !sendReasoningDelta(streamResp.Delta) {
-		//	return false
+		//	sr.Stop(streamErr)
+		//	return
 		//}
 
 		//case "response.reasoning_text.done":
 
 		case "response.reasoning_summary_text.delta":
 			if !sendReasoningSummaryDelta(streamResp.Delta) {
-				return false
+				sr.Stop(streamErr)
+				return
 			}
 
 		case "response.reasoning_summary_text.done":
@@ -349,12 +353,14 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 		//	delta := stringDeltaFromPrefix(prev, next)
 		//	reasoningSummaryTextByKey[key] = next
 		//	if !sendReasoningSummaryDelta(delta) {
-		//		return false
+		//		sr.Stop(streamErr)
+		//		return
 		//	}
 
 		case "response.output_text.delta":
 			if !sendStartIfNeeded() {
-				return false
+				sr.Stop(streamErr)
+				return
 			}
 
 			if streamResp.Delta != "" {
@@ -376,7 +382,8 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 					},
 				}
 				if !sendChatChunk(chunk) {
-					return false
+					sr.Stop(streamErr)
+					return
 				}
 			}
 
@@ -414,7 +421,8 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 			}
 
 			if !sendToolCallDelta(callID, name, argsDelta) {
-				return false
+				sr.Stop(streamErr)
+				return
 			}
 
 		case "response.function_call_arguments.delta":
@@ -428,7 +436,8 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 			}
 			toolCallArgsByID[callID] += streamResp.Delta
 			if !sendToolCallDelta(callID, "", streamResp.Delta) {
-				return false
+				sr.Stop(streamErr)
+				return
 			}
 
 		case "response.function_call_arguments.done":
@@ -467,7 +476,8 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 			}
 
 			if !sendStartIfNeeded() {
-				return false
+				sr.Stop(streamErr)
+				return
 			}
 			if !sentStop {
 				if info.RelayFormat == types.RelayFormatClaude && info.ClaudeConvertInfo != nil {
@@ -479,7 +489,8 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 				}
 				stop := helper.GenerateStopResponse(responseId, createAt, model, finishReason)
 				if !sendChatChunk(stop) {
-					return false
+					sr.Stop(streamErr)
+					return
 				}
 				sentStop = true
 			}
@@ -488,16 +499,16 @@ func OaiResponsesToChatStreamHandler(c *gin.Context, info *relaycommon.RelayInfo
 			if streamResp.Response != nil {
 				if oaiErr := streamResp.Response.GetOpenAIError(); oaiErr != nil && oaiErr.Type != "" {
 					streamErr = types.WithOpenAIError(*oaiErr, http.StatusInternalServerError)
-					return false
+					sr.Stop(streamErr)
+					return
 				}
 			}
 			streamErr = types.NewOpenAIError(fmt.Errorf("responses stream error: %s", streamResp.Type), types.ErrorCodeBadResponse, http.StatusInternalServerError)
-			return false
+			sr.Stop(streamErr)
+			return
 
 		default:
 		}
-
-		return true
 	})
 
 	if streamErr != nil {

relay/channel/openai/relay-openai.go

@@ -126,11 +126,11 @@ func OaiStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Re
 	// 检查是否为音频模型
 	isAudioModel := strings.Contains(strings.ToLower(model), "audio")
 
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 		if lastStreamData != "" {
-			err := HandleStreamFormat(c, info, lastStreamData, info.ChannelSetting.ForceFormat, info.ChannelSetting.ThinkingToContent)
-			if err != nil {
+			if err := HandleStreamFormat(c, info, lastStreamData, info.ChannelSetting.ForceFormat, info.ChannelSetting.ThinkingToContent); err != nil {
 				common.SysLog("error handling stream format: " + err.Error())
+				sr.Error(err)
 			}
 		}
 		if len(data) > 0 {
@@ -142,7 +142,6 @@ func OaiStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Re
 			lastStreamData = data
 			streamItems = append(streamItems, data)
 		}
-		return true
 	})
 
 	// 对音频模型，从倒数第二个stream data中提取usage信息

relay/channel/openai/relay_responses.go

@@ -79,55 +79,55 @@ func OaiResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
 	var usage = &dto.Usage{}
 	var responseTextBuilder strings.Builder
 
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 
 		// 检查当前数据是否包含 completed 状态和 usage 信息
 		var streamResponse dto.ResponsesStreamResponse
-		if err := common.UnmarshalJsonStr(data, &streamResponse); err == nil {
-			sendResponsesStreamData(c, streamResponse, data)
-			switch streamResponse.Type {
-			case "response.completed":
-				if streamResponse.Response != nil {
-					if streamResponse.Response.Usage != nil {
-						if streamResponse.Response.Usage.InputTokens != 0 {
-							usage.PromptTokens = streamResponse.Response.Usage.InputTokens
-						}
-						if streamResponse.Response.Usage.OutputTokens != 0 {
-							usage.CompletionTokens = streamResponse.Response.Usage.OutputTokens
-						}
-						if streamResponse.Response.Usage.TotalTokens != 0 {
-							usage.TotalTokens = streamResponse.Response.Usage.TotalTokens
-						}
-						if streamResponse.Response.Usage.InputTokensDetails != nil {
-							usage.PromptTokensDetails.CachedTokens = streamResponse.Response.Usage.InputTokensDetails.CachedTokens
-						}
+		if err := common.UnmarshalJsonStr(data, &streamResponse); err != nil {
+			logger.LogError(c, "failed to unmarshal stream response: "+err.Error())
+			sr.Error(err)
+			return
+		}
+		sendResponsesStreamData(c, streamResponse, data)
+		switch streamResponse.Type {
+		case "response.completed":
+			if streamResponse.Response != nil {
+				if streamResponse.Response.Usage != nil {
+					if streamResponse.Response.Usage.InputTokens != 0 {
+						usage.PromptTokens = streamResponse.Response.Usage.InputTokens
 					}
-					if streamResponse.Response.HasImageGenerationCall() {
-						c.Set("image_generation_call", true)
-						c.Set("image_generation_call_quality", streamResponse.Response.GetQuality())
-						c.Set("image_generation_call_size", streamResponse.Response.GetSize())
+					if streamResponse.Response.Usage.OutputTokens != 0 {
+						usage.CompletionTokens = streamResponse.Response.Usage.OutputTokens
+					}
+					if streamResponse.Response.Usage.TotalTokens != 0 {
+						usage.TotalTokens = streamResponse.Response.Usage.TotalTokens
+					}
+					if streamResponse.Response.Usage.InputTokensDetails != nil {
+						usage.PromptTokensDetails.CachedTokens = streamResponse.Response.Usage.InputTokensDetails.CachedTokens
 					}
 				}
-			case "response.output_text.delta":
-				// 处理输出文本
-				responseTextBuilder.WriteString(streamResponse.Delta)
-			case dto.ResponsesOutputTypeItemDone:
-				// 函数调用处理
-				if streamResponse.Item != nil {
-					switch streamResponse.Item.Type {
-					case dto.BuildInCallWebSearchCall:
-						if info != nil && info.ResponsesUsageInfo != nil && info.ResponsesUsageInfo.BuiltInTools != nil {
-							if webSearchTool, exists := info.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists && webSearchTool != nil {
-								webSearchTool.CallCount++
-							}
+				if streamResponse.Response.HasImageGenerationCall() {
+					c.Set("image_generation_call", true)
+					c.Set("image_generation_call_quality", streamResponse.Response.GetQuality())
+					c.Set("image_generation_call_size", streamResponse.Response.GetSize())
+				}
+			}
+		case "response.output_text.delta":
+			// 处理输出文本
+			responseTextBuilder.WriteString(streamResponse.Delta)
+		case dto.ResponsesOutputTypeItemDone:
+			// 函数调用处理
+			if streamResponse.Item != nil {
+				switch streamResponse.Item.Type {
+				case dto.BuildInCallWebSearchCall:
+					if info != nil && info.ResponsesUsageInfo != nil && info.ResponsesUsageInfo.BuiltInTools != nil {
+						if webSearchTool, exists := info.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists && webSearchTool != nil {
+							webSearchTool.CallCount++
 						}
 					}
 				}
 			}
-		} else {
-			logger.LogError(c, "failed to unmarshal stream response: "+err.Error())
 		}
-		return true
 	})
 
 	if usage.CompletionTokens == 0 {

relay/channel/task/taskcommon/helpers.go

@@ -17,6 +17,8 @@ func UnmarshalMetadata(metadata map[string]any, target any) error {
 	if metadata == nil {
 		return nil
 	}
+	// Prevent metadata from overriding model fields to avoid billing bypass.
+	delete(metadata, "model")
 	metaBytes, err := common.Marshal(metadata)
 	if err != nil {
 		return fmt.Errorf("marshal metadata failed: %w", err)

relay/channel/xai/text.go

@@ -43,12 +43,12 @@ func xAIStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Re
 
 	helper.SetEventStreamHeaders(c)
 
-	helper.StreamScannerHandler(c, resp, info, func(data string) bool {
+	helper.StreamScannerHandler(c, resp, info, func(data string, sr *helper.StreamResult) {
 		var xAIResp *dto.ChatCompletionsStreamResponse
-		err := common.UnmarshalJsonStr(data, &xAIResp)
-		if err != nil {
+		if err := common.UnmarshalJsonStr(data, &xAIResp); err != nil {
 			common.SysLog("error unmarshalling stream response: " + err.Error())
-			return true
+			sr.Error(err)
+			return
 		}
 
 		// 把 xAI 的usage转换为 OpenAI 的usage
@@ -61,11 +61,10 @@ func xAIStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http.Re
 
 		openaiResponse := streamResponseXAI2OpenAI(xAIResp, usage)
 		_ = openai.ProcessStreamResponse(*openaiResponse, &responseTextBuilder, &toolCount)
-		err = helper.ObjectData(c, openaiResponse)
-		if err != nil {
+		if err := helper.ObjectData(c, openaiResponse); err != nil {
 			common.SysLog(err.Error())
+			sr.Error(err)
 		}
-		return true
 	})
 
 	if !containStreamUsage {

relay/claude_handler.go

@@ -122,7 +122,7 @@ func ClaudeHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
 			return newApiErr
 		}
 
-		service.PostClaudeConsumeQuota(c, info, usage)
+		service.PostTextConsumeQuota(c, info, usage, nil)
 		return nil
 	}
 
@@ -190,6 +190,6 @@ func ClaudeHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
 		return newAPIError
 	}
 
-	service.PostClaudeConsumeQuota(c, info, usage.(*dto.Usage))
+	service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), nil)
 	return nil
 }

relay/common/relay_info.go

@@ -162,6 +162,8 @@ type RelayInfo struct {
 	// 若为空，调用 GetFinalRequestRelayFormat 会回退到 RequestConversionChain 的最后一项或 RelayFormat。
 	FinalRequestRelayFormat types.RelayFormat
 
+	StreamStatus *StreamStatus
+
 	ThinkingContentInfo
 	TokenCountMeta
 	*ClaudeConvertInfo
@@ -338,15 +340,10 @@ func GenRelayInfoClaude(c *gin.Context, request dto.Request) *RelayInfo {
 	info.ClaudeConvertInfo = &ClaudeConvertInfo{
 		LastMessagesType: LastMessageTypeNone,
 	}
-	info.IsClaudeBetaQuery = c.Query("beta") == "true" || isClaudeBetaForced(c)
+	info.IsClaudeBetaQuery = c.Query("beta") == "true"
 	return info
 }
 
-func isClaudeBetaForced(c *gin.Context) bool {
-	channelOtherSettings, ok := common.GetContextKeyType[dto.ChannelOtherSettings](c, constant.ContextKeyChannelOtherSetting)
-	return ok && channelOtherSettings.ClaudeBetaQuery
-}
-
 func GenRelayInfoRerank(c *gin.Context, request *dto.RerankRequest) *RelayInfo {
 	info := genBaseRelayInfo(c, request)
 	info.RelayMode = relayconstant.RelayModeRerank

relay/common/stream_status.go

@@ -0,0 +1,112 @@
+package common
+
+import (
+	"fmt"
+	"strings"
+	"sync"
+	"time"
+)
+
+type StreamEndReason string
+
+const (
+	StreamEndReasonNone        StreamEndReason = ""
+	StreamEndReasonDone        StreamEndReason = "done"
+	StreamEndReasonTimeout     StreamEndReason = "timeout"
+	StreamEndReasonClientGone  StreamEndReason = "client_gone"
+	StreamEndReasonScannerErr  StreamEndReason = "scanner_error"
+	StreamEndReasonHandlerStop StreamEndReason = "handler_stop"
+	StreamEndReasonEOF         StreamEndReason = "eof"
+	StreamEndReasonPanic       StreamEndReason = "panic"
+	StreamEndReasonPingFail    StreamEndReason = "ping_fail"
+)
+
+const maxStreamErrorEntries = 20
+
+type StreamErrorEntry struct {
+	Message   string
+	Timestamp time.Time
+}
+
+type StreamStatus struct {
+	EndReason  StreamEndReason
+	EndError   error
+	endOnce    sync.Once
+
+	mu         sync.Mutex
+	Errors     []StreamErrorEntry
+	ErrorCount int
+}
+
+func NewStreamStatus() *StreamStatus {
+	return &StreamStatus{}
+}
+
+func (s *StreamStatus) SetEndReason(reason StreamEndReason, err error) {
+	if s == nil {
+		return
+	}
+	s.endOnce.Do(func() {
+		s.EndReason = reason
+		s.EndError = err
+	})
+}
+
+func (s *StreamStatus) RecordError(msg string) {
+	if s == nil {
+		return
+	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	s.ErrorCount++
+	if len(s.Errors) < maxStreamErrorEntries {
+		s.Errors = append(s.Errors, StreamErrorEntry{
+			Message:   msg,
+			Timestamp: time.Now(),
+		})
+	}
+}
+
+func (s *StreamStatus) HasErrors() bool {
+	if s == nil {
+		return false
+	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return s.ErrorCount > 0
+}
+
+func (s *StreamStatus) TotalErrorCount() int {
+	if s == nil {
+		return 0
+	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return s.ErrorCount
+}
+
+func (s *StreamStatus) IsNormalEnd() bool {
+	if s == nil {
+		return true
+	}
+	return s.EndReason == StreamEndReasonDone ||
+		s.EndReason == StreamEndReasonEOF ||
+		s.EndReason == StreamEndReasonHandlerStop
+}
+
+func (s *StreamStatus) Summary() string {
+	if s == nil {
+		return "StreamStatus<nil>"
+	}
+	b := &strings.Builder{}
+	fmt.Fprintf(b, "reason=%s", s.EndReason)
+	if s.EndError != nil {
+		fmt.Fprintf(b, " end_error=%q", s.EndError.Error())
+	}
+	s.mu.Lock()
+	if s.ErrorCount > 0 {
+		fmt.Fprintf(b, " soft_errors=%d", s.ErrorCount)
+	}
+	s.mu.Unlock()
+	return b.String()
+}

relay/common/stream_status_test.go

@@ -0,0 +1,182 @@
+package common
+
+import (
+	"fmt"
+	"sync"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestStreamStatus_SetEndReason_FirstWins(t *testing.T) {
+	t.Parallel()
+	s := NewStreamStatus()
+
+	s.SetEndReason(StreamEndReasonDone, nil)
+	s.SetEndReason(StreamEndReasonTimeout, nil)
+	s.SetEndReason(StreamEndReasonClientGone, fmt.Errorf("context canceled"))
+
+	assert.Equal(t, StreamEndReasonDone, s.EndReason)
+	assert.Nil(t, s.EndError)
+}
+
+func TestStreamStatus_SetEndReason_WithError(t *testing.T) {
+	t.Parallel()
+	s := NewStreamStatus()
+
+	expectedErr := fmt.Errorf("read: connection reset")
+	s.SetEndReason(StreamEndReasonScannerErr, expectedErr)
+
+	assert.Equal(t, StreamEndReasonScannerErr, s.EndReason)
+	assert.Equal(t, expectedErr, s.EndError)
+}
+
+func TestStreamStatus_SetEndReason_NilSafe(t *testing.T) {
+	t.Parallel()
+	var s *StreamStatus
+	s.SetEndReason(StreamEndReasonDone, nil)
+}
+
+func TestStreamStatus_SetEndReason_Concurrent(t *testing.T) {
+	t.Parallel()
+	s := NewStreamStatus()
+
+	reasons := []StreamEndReason{
+		StreamEndReasonDone,
+		StreamEndReasonTimeout,
+		StreamEndReasonClientGone,
+		StreamEndReasonScannerErr,
+		StreamEndReasonHandlerStop,
+		StreamEndReasonEOF,
+		StreamEndReasonPanic,
+		StreamEndReasonPingFail,
+	}
+
+	var wg sync.WaitGroup
+	for _, r := range reasons {
+		wg.Add(1)
+		go func(reason StreamEndReason) {
+			defer wg.Done()
+			s.SetEndReason(reason, nil)
+		}(r)
+	}
+	wg.Wait()
+
+	assert.NotEqual(t, StreamEndReasonNone, s.EndReason)
+}
+
+func TestStreamStatus_RecordError_Basic(t *testing.T) {
+	t.Parallel()
+	s := NewStreamStatus()
+
+	s.RecordError("bad json")
+	s.RecordError("another bad json")
+	s.RecordError("client gone")
+
+	assert.True(t, s.HasErrors())
+	assert.Equal(t, 3, s.TotalErrorCount())
+	assert.Len(t, s.Errors, 3)
+}
+
+func TestStreamStatus_RecordError_CapAtMax(t *testing.T) {
+	t.Parallel()
+	s := NewStreamStatus()
+
+	for i := 0; i < 30; i++ {
+		s.RecordError(fmt.Sprintf("error_%d", i))
+	}
+
+	assert.Equal(t, maxStreamErrorEntries, len(s.Errors))
+	assert.Equal(t, 30, s.TotalErrorCount())
+}
+
+func TestStreamStatus_RecordError_NilSafe(t *testing.T) {
+	t.Parallel()
+	var s *StreamStatus
+	s.RecordError("should not panic")
+}
+
+func TestStreamStatus_RecordError_Concurrent(t *testing.T) {
+	t.Parallel()
+	s := NewStreamStatus()
+
+	var wg sync.WaitGroup
+	for i := 0; i < 100; i++ {
+		wg.Add(1)
+		go func(idx int) {
+			defer wg.Done()
+			s.RecordError(fmt.Sprintf("error_%d", idx))
+		}(i)
+	}
+	wg.Wait()
+
+	assert.Equal(t, 100, s.TotalErrorCount())
+	assert.LessOrEqual(t, len(s.Errors), maxStreamErrorEntries)
+}
+
+func TestStreamStatus_HasErrors_Empty(t *testing.T) {
+	t.Parallel()
+	s := NewStreamStatus()
+	assert.False(t, s.HasErrors())
+	assert.Equal(t, 0, s.TotalErrorCount())
+}
+
+func TestStreamStatus_HasErrors_NilSafe(t *testing.T) {
+	t.Parallel()
+	var s *StreamStatus
+	assert.False(t, s.HasErrors())
+	assert.Equal(t, 0, s.TotalErrorCount())
+}
+
+func TestStreamStatus_IsNormalEnd(t *testing.T) {
+	t.Parallel()
+	tests := []struct {
+		reason StreamEndReason
+		normal bool
+	}{
+		{StreamEndReasonDone, true},
+		{StreamEndReasonEOF, true},
+		{StreamEndReasonHandlerStop, true},
+		{StreamEndReasonTimeout, false},
+		{StreamEndReasonClientGone, false},
+		{StreamEndReasonScannerErr, false},
+		{StreamEndReasonPanic, false},
+		{StreamEndReasonPingFail, false},
+		{StreamEndReasonNone, false},
+	}
+	for _, tt := range tests {
+		s := NewStreamStatus()
+		s.SetEndReason(tt.reason, nil)
+		assert.Equal(t, tt.normal, s.IsNormalEnd(), "reason=%s", tt.reason)
+	}
+}
+
+func TestStreamStatus_IsNormalEnd_NilSafe(t *testing.T) {
+	t.Parallel()
+	var s *StreamStatus
+	assert.True(t, s.IsNormalEnd())
+}
+
+func TestStreamStatus_Summary(t *testing.T) {
+	t.Parallel()
+
+	s := NewStreamStatus()
+	s.SetEndReason(StreamEndReasonDone, nil)
+	summary := s.Summary()
+	assert.Contains(t, summary, "reason=done")
+	assert.NotContains(t, summary, "soft_errors")
+
+	s2 := NewStreamStatus()
+	s2.SetEndReason(StreamEndReasonTimeout, nil)
+	s2.RecordError("bad json")
+	s2.RecordError("write failed")
+	summary2 := s2.Summary()
+	assert.Contains(t, summary2, "reason=timeout")
+	assert.Contains(t, summary2, "soft_errors=2")
+}
+
+func TestStreamStatus_Summary_NilSafe(t *testing.T) {
+	t.Parallel()
+	var s *StreamStatus
+	assert.Equal(t, "StreamStatus<nil>", s.Summary())
+}

relay/compatible_handler.go

@@ -6,25 +6,20 @@ import (
 	"io"
 	"net/http"
 	"strings"
-	"time"
 
 	"github.com/QuantumNous/new-api/common"
 	"github.com/QuantumNous/new-api/constant"
 	"github.com/QuantumNous/new-api/dto"
 	"github.com/QuantumNous/new-api/logger"
-	"github.com/QuantumNous/new-api/model"
 	relaycommon "github.com/QuantumNous/new-api/relay/common"
 	relayconstant "github.com/QuantumNous/new-api/relay/constant"
 	"github.com/QuantumNous/new-api/relay/helper"
 	"github.com/QuantumNous/new-api/service"
 	"github.com/QuantumNous/new-api/setting/model_setting"
-	"github.com/QuantumNous/new-api/setting/operation_setting"
 	"github.com/QuantumNous/new-api/setting/ratio_setting"
 	"github.com/QuantumNous/new-api/types"
 	"github.com/samber/lo"
 
-	"github.com/shopspring/decimal"
-
 	"github.com/gin-gonic/gin"
 )
 
@@ -93,7 +88,7 @@ func TextHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *types
 		if containAudioTokens && containsAudioRatios {
 			service.PostAudioConsumeQuota(c, info, usage, "")
 		} else {
-			postConsumeQuota(c, info, usage)
+			service.PostTextConsumeQuota(c, info, usage, nil)
 		}
 		return nil
 	}
@@ -216,293 +211,7 @@ func TextHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *types
 	if containAudioTokens && containsAudioRatios {
 		service.PostAudioConsumeQuota(c, info, usage.(*dto.Usage), "")
 	} else {
-		postConsumeQuota(c, info, usage.(*dto.Usage))
+		service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), nil)
 	}
 	return nil
 }
-
-func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.Usage, extraContent ...string) {
-	originUsage := usage
-	if usage == nil {
-		usage = &dto.Usage{
-			PromptTokens:     relayInfo.GetEstimatePromptTokens(),
-			CompletionTokens: 0,
-			TotalTokens:      relayInfo.GetEstimatePromptTokens(),
-		}
-		extraContent = append(extraContent, "上游无计费信息")
-	}
-
-	if originUsage != nil {
-		service.ObserveChannelAffinityUsageCacheByRelayFormat(ctx, usage, relayInfo.GetFinalRequestRelayFormat())
-	}
-
-	adminRejectReason := common.GetContextKeyString(ctx, constant.ContextKeyAdminRejectReason)
-
-	useTimeSeconds := time.Now().Unix() - relayInfo.StartTime.Unix()
-	promptTokens := usage.PromptTokens
-	cacheTokens := usage.PromptTokensDetails.CachedTokens
-	imageTokens := usage.PromptTokensDetails.ImageTokens
-	audioTokens := usage.PromptTokensDetails.AudioTokens
-	completionTokens := usage.CompletionTokens
-	cachedCreationTokens := usage.PromptTokensDetails.CachedCreationTokens
-
-	modelName := relayInfo.OriginModelName
-
-	tokenName := ctx.GetString("token_name")
-	completionRatio := relayInfo.PriceData.CompletionRatio
-	cacheRatio := relayInfo.PriceData.CacheRatio
-	imageRatio := relayInfo.PriceData.ImageRatio
-	modelRatio := relayInfo.PriceData.ModelRatio
-	groupRatio := relayInfo.PriceData.GroupRatioInfo.GroupRatio
-	modelPrice := relayInfo.PriceData.ModelPrice
-	cachedCreationRatio := relayInfo.PriceData.CacheCreationRatio
-
-	// Convert values to decimal for precise calculation
-	dPromptTokens := decimal.NewFromInt(int64(promptTokens))
-	dCacheTokens := decimal.NewFromInt(int64(cacheTokens))
-	dImageTokens := decimal.NewFromInt(int64(imageTokens))
-	dAudioTokens := decimal.NewFromInt(int64(audioTokens))
-	dCompletionTokens := decimal.NewFromInt(int64(completionTokens))
-	dCachedCreationTokens := decimal.NewFromInt(int64(cachedCreationTokens))
-	dCompletionRatio := decimal.NewFromFloat(completionRatio)
-	dCacheRatio := decimal.NewFromFloat(cacheRatio)
-	dImageRatio := decimal.NewFromFloat(imageRatio)
-	dModelRatio := decimal.NewFromFloat(modelRatio)
-	dGroupRatio := decimal.NewFromFloat(groupRatio)
-	dModelPrice := decimal.NewFromFloat(modelPrice)
-	dCachedCreationRatio := decimal.NewFromFloat(cachedCreationRatio)
-	dQuotaPerUnit := decimal.NewFromFloat(common.QuotaPerUnit)
-
-	ratio := dModelRatio.Mul(dGroupRatio)
-
-	// openai web search 工具计费
-	var dWebSearchQuota decimal.Decimal
-	var webSearchPrice float64
-	// response api 格式工具计费
-	if relayInfo.ResponsesUsageInfo != nil {
-		if webSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists && webSearchTool.CallCount > 0 {
-			// 计算 web search 调用的配额 (配额 = 价格 * 调用次数 / 1000 * 分组倍率)
-			webSearchPrice = operation_setting.GetWebSearchPricePerThousand(modelName, webSearchTool.SearchContextSize)
-			dWebSearchQuota = decimal.NewFromFloat(webSearchPrice).
-				Mul(decimal.NewFromInt(int64(webSearchTool.CallCount))).
-				Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
-			extraContent = append(extraContent, fmt.Sprintf("Web Search 调用 %d 次，上下文大小 %s，调用花费 %s",
-				webSearchTool.CallCount, webSearchTool.SearchContextSize, dWebSearchQuota.String()))
-		}
-	} else if strings.HasSuffix(modelName, "search-preview") {
-		// search-preview 模型不支持 response api
-		searchContextSize := ctx.GetString("chat_completion_web_search_context_size")
-		if searchContextSize == "" {
-			searchContextSize = "medium"
-		}
-		webSearchPrice = operation_setting.GetWebSearchPricePerThousand(modelName, searchContextSize)
-		dWebSearchQuota = decimal.NewFromFloat(webSearchPrice).
-			Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
-		extraContent = append(extraContent, fmt.Sprintf("Web Search 调用 1 次，上下文大小 %s，调用花费 %s",
-			searchContextSize, dWebSearchQuota.String()))
-	}
-	// claude web search tool 计费
-	var dClaudeWebSearchQuota decimal.Decimal
-	var claudeWebSearchPrice float64
-	claudeWebSearchCallCount := ctx.GetInt("claude_web_search_requests")
-	if claudeWebSearchCallCount > 0 {
-		claudeWebSearchPrice = operation_setting.GetClaudeWebSearchPricePerThousand()
-		dClaudeWebSearchQuota = decimal.NewFromFloat(claudeWebSearchPrice).
-			Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit).Mul(decimal.NewFromInt(int64(claudeWebSearchCallCount)))
-		extraContent = append(extraContent, fmt.Sprintf("Claude Web Search 调用 %d 次，调用花费 %s",
-			claudeWebSearchCallCount, dClaudeWebSearchQuota.String()))
-	}
-	// file search tool 计费
-	var dFileSearchQuota decimal.Decimal
-	var fileSearchPrice float64
-	if relayInfo.ResponsesUsageInfo != nil {
-		if fileSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolFileSearch]; exists && fileSearchTool.CallCount > 0 {
-			fileSearchPrice = operation_setting.GetFileSearchPricePerThousand()
-			dFileSearchQuota = decimal.NewFromFloat(fileSearchPrice).
-				Mul(decimal.NewFromInt(int64(fileSearchTool.CallCount))).
-				Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
-			extraContent = append(extraContent, fmt.Sprintf("File Search 调用 %d 次，调用花费 %s",
-				fileSearchTool.CallCount, dFileSearchQuota.String()))
-		}
-	}
-	var dImageGenerationCallQuota decimal.Decimal
-	var imageGenerationCallPrice float64
-	if ctx.GetBool("image_generation_call") {
-		imageGenerationCallPrice = operation_setting.GetGPTImage1PriceOnceCall(ctx.GetString("image_generation_call_quality"), ctx.GetString("image_generation_call_size"))
-		dImageGenerationCallQuota = decimal.NewFromFloat(imageGenerationCallPrice).Mul(dGroupRatio).Mul(dQuotaPerUnit)
-		extraContent = append(extraContent, fmt.Sprintf("Image Generation Call 花费 %s", dImageGenerationCallQuota.String()))
-	}
-
-	var quotaCalculateDecimal decimal.Decimal
-
-	var audioInputQuota decimal.Decimal
-	var audioInputPrice float64
-	isClaudeUsageSemantic := relayInfo.GetFinalRequestRelayFormat() == types.RelayFormatClaude
-	if !relayInfo.PriceData.UsePrice {
-		baseTokens := dPromptTokens
-		// 减去 cached tokens
-		// Anthropic API 的 input_tokens 已经不包含缓存 tokens，不需要减去
-		// OpenAI/OpenRouter 等 API 的 prompt_tokens 包含缓存 tokens，需要减去
-		var cachedTokensWithRatio decimal.Decimal
-		if !dCacheTokens.IsZero() {
-			if !isClaudeUsageSemantic {
-				baseTokens = baseTokens.Sub(dCacheTokens)
-			}
-			cachedTokensWithRatio = dCacheTokens.Mul(dCacheRatio)
-		}
-		var dCachedCreationTokensWithRatio decimal.Decimal
-		if !dCachedCreationTokens.IsZero() {
-			if !isClaudeUsageSemantic {
-				baseTokens = baseTokens.Sub(dCachedCreationTokens)
-			}
-			dCachedCreationTokensWithRatio = dCachedCreationTokens.Mul(dCachedCreationRatio)
-		}
-
-		// 减去 image tokens
-		var imageTokensWithRatio decimal.Decimal
-		if !dImageTokens.IsZero() {
-			baseTokens = baseTokens.Sub(dImageTokens)
-			imageTokensWithRatio = dImageTokens.Mul(dImageRatio)
-		}
-
-		// 减去 Gemini audio tokens
-		if !dAudioTokens.IsZero() {
-			audioInputPrice = operation_setting.GetGeminiInputAudioPricePerMillionTokens(modelName)
-			if audioInputPrice > 0 {
-				// 重新计算 base tokens
-				baseTokens = baseTokens.Sub(dAudioTokens)
-				audioInputQuota = decimal.NewFromFloat(audioInputPrice).Div(decimal.NewFromInt(1000000)).Mul(dAudioTokens).Mul(dGroupRatio).Mul(dQuotaPerUnit)
-				extraContent = append(extraContent, fmt.Sprintf("Audio Input 花费 %s", audioInputQuota.String()))
-			}
-		}
-		promptQuota := baseTokens.Add(cachedTokensWithRatio).
-			Add(imageTokensWithRatio).
-			Add(dCachedCreationTokensWithRatio)
-
-		completionQuota := dCompletionTokens.Mul(dCompletionRatio)
-
-		quotaCalculateDecimal = promptQuota.Add(completionQuota).Mul(ratio)
-
-		if !ratio.IsZero() && quotaCalculateDecimal.LessThanOrEqual(decimal.Zero) {
-			quotaCalculateDecimal = decimal.NewFromInt(1)
-		}
-	} else {
-		quotaCalculateDecimal = dModelPrice.Mul(dQuotaPerUnit).Mul(dGroupRatio)
-	}
-	// 添加 responses tools call 调用的配额
-	quotaCalculateDecimal = quotaCalculateDecimal.Add(dWebSearchQuota)
-	quotaCalculateDecimal = quotaCalculateDecimal.Add(dFileSearchQuota)
-	// 添加 audio input 独立计费
-	quotaCalculateDecimal = quotaCalculateDecimal.Add(audioInputQuota)
-	// 添加 image generation call 计费
-	quotaCalculateDecimal = quotaCalculateDecimal.Add(dImageGenerationCallQuota)
-
-	if len(relayInfo.PriceData.OtherRatios) > 0 {
-		for key, otherRatio := range relayInfo.PriceData.OtherRatios {
-			dOtherRatio := decimal.NewFromFloat(otherRatio)
-			quotaCalculateDecimal = quotaCalculateDecimal.Mul(dOtherRatio)
-			extraContent = append(extraContent, fmt.Sprintf("其他倍率 %s: %f", key, otherRatio))
-		}
-	}
-
-	quota := int(quotaCalculateDecimal.Round(0).IntPart())
-	totalTokens := promptTokens + completionTokens
-
-	//var logContent string
-
-	// record all the consume log even if quota is 0
-	if totalTokens == 0 {
-		// in this case, must be some error happened
-		// we cannot just return, because we may have to return the pre-consumed quota
-		quota = 0
-		extraContent = append(extraContent, "上游没有返回计费信息，无法扣费（可能是上游超时）")
-		logger.LogError(ctx, fmt.Sprintf("total tokens is 0, cannot consume quota, userId %d, channelId %d, "+
-			"tokenId %d, model %s， pre-consumed quota %d", relayInfo.UserId, relayInfo.ChannelId, relayInfo.TokenId, modelName, relayInfo.FinalPreConsumedQuota))
-	} else {
-		if !ratio.IsZero() && quota == 0 {
-			quota = 1
-		}
-		model.UpdateUserUsedQuotaAndRequestCount(relayInfo.UserId, quota)
-		model.UpdateChannelUsedQuota(relayInfo.ChannelId, quota)
-	}
-
-	if err := service.SettleBilling(ctx, relayInfo, quota); err != nil {
-		logger.LogError(ctx, "error settling billing: "+err.Error())
-	}
-
-	logModel := modelName
-	if strings.HasPrefix(logModel, "gpt-4-gizmo") {
-		logModel = "gpt-4-gizmo-*"
-		extraContent = append(extraContent, fmt.Sprintf("模型 %s", modelName))
-	}
-	if strings.HasPrefix(logModel, "gpt-4o-gizmo") {
-		logModel = "gpt-4o-gizmo-*"
-		extraContent = append(extraContent, fmt.Sprintf("模型 %s", modelName))
-	}
-	logContent := strings.Join(extraContent, ", ")
-	other := service.GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, cacheTokens, cacheRatio, modelPrice, relayInfo.PriceData.GroupRatioInfo.GroupSpecialRatio)
-	if adminRejectReason != "" {
-		other["reject_reason"] = adminRejectReason
-	}
-	// For chat-based calls to the Claude model, tagging is required. Using Claude's rendering logs, the two approaches handle input rendering differently.
-	if isClaudeUsageSemantic {
-		other["claude"] = true
-		other["usage_semantic"] = "anthropic"
-	}
-	if imageTokens != 0 {
-		other["image"] = true
-		other["image_ratio"] = imageRatio
-		other["image_output"] = imageTokens
-	}
-	if cachedCreationTokens != 0 {
-		other["cache_creation_tokens"] = cachedCreationTokens
-		other["cache_creation_ratio"] = cachedCreationRatio
-	}
-	if !dWebSearchQuota.IsZero() {
-		if relayInfo.ResponsesUsageInfo != nil {
-			if webSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists {
-				other["web_search"] = true
-				other["web_search_call_count"] = webSearchTool.CallCount
-				other["web_search_price"] = webSearchPrice
-			}
-		} else if strings.HasSuffix(modelName, "search-preview") {
-			other["web_search"] = true
-			other["web_search_call_count"] = 1
-			other["web_search_price"] = webSearchPrice
-		}
-	} else if !dClaudeWebSearchQuota.IsZero() {
-		other["web_search"] = true
-		other["web_search_call_count"] = claudeWebSearchCallCount
-		other["web_search_price"] = claudeWebSearchPrice
-	}
-	if !dFileSearchQuota.IsZero() && relayInfo.ResponsesUsageInfo != nil {
-		if fileSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolFileSearch]; exists {
-			other["file_search"] = true
-			other["file_search_call_count"] = fileSearchTool.CallCount
-			other["file_search_price"] = fileSearchPrice
-		}
-	}
-	if !audioInputQuota.IsZero() {
-		other["audio_input_seperate_price"] = true
-		other["audio_input_token_count"] = audioTokens
-		other["audio_input_price"] = audioInputPrice
-	}
-	if !dImageGenerationCallQuota.IsZero() {
-		other["image_generation_call"] = true
-		other["image_generation_call_price"] = imageGenerationCallPrice
-	}
-	model.RecordConsumeLog(ctx, relayInfo.UserId, model.RecordConsumeLogParams{
-		ChannelId:        relayInfo.ChannelId,
-		PromptTokens:     promptTokens,
-		CompletionTokens: completionTokens,
-		ModelName:        logModel,
-		TokenName:        tokenName,
-		Quota:            quota,
-		Content:          logContent,
-		TokenId:          relayInfo.TokenId,
-		UseTimeSeconds:   int(useTimeSeconds),
-		IsStream:         relayInfo.IsStream,
-		Group:            relayInfo.UsingGroup,
-		Other:            other,
-	})
-}

relay/embedding_handler.go

@@ -82,6 +82,6 @@ func EmbeddingHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *
 		service.ResetStatusCode(newAPIError, statusCodeMappingStr)
 		return newAPIError
 	}
-	postConsumeQuota(c, info, usage.(*dto.Usage))
+	service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), nil)
 	return nil
 }

relay/gemini_handler.go

@@ -194,7 +194,7 @@ func GeminiHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
 		return openaiErr
 	}
 
-	postConsumeQuota(c, info, usage.(*dto.Usage))
+	service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), nil)
 	return nil
 }
 
@@ -288,6 +288,6 @@ func GeminiEmbeddingHandler(c *gin.Context, info *relaycommon.RelayInfo) (newAPI
 		return openaiErr
 	}
 
-	postConsumeQuota(c, info, usage.(*dto.Usage))
+	service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), nil)
 	return nil
 }

relay/helper/stream_result.go

@@ -0,0 +1,52 @@
+package helper
+
+import (
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+)
+
+// StreamResult is passed to each dataHandler invocation, providing methods
+// to record soft errors, signal fatal stops, or mark normal completion.
+// StreamScannerHandler checks IsStopped() after each callback invocation.
+type StreamResult struct {
+	status  *relaycommon.StreamStatus
+	stopped bool
+}
+
+func newStreamResult(status *relaycommon.StreamStatus) *StreamResult {
+	return &StreamResult{status: status}
+}
+
+// Error records a soft error. The stream continues processing.
+// Can be called multiple times per chunk.
+func (r *StreamResult) Error(err error) {
+	if err == nil {
+		return
+	}
+	r.status.RecordError(err.Error())
+}
+
+// Stop records a fatal error and marks the stream to stop after this chunk.
+func (r *StreamResult) Stop(err error) {
+	if err != nil {
+		r.status.RecordError(err.Error())
+	}
+	r.status.SetEndReason(relaycommon.StreamEndReasonHandlerStop, err)
+	r.stopped = true
+}
+
+// Done signals that the handler has finished processing normally
+// (e.g., Dify "message_end"). The stream stops after this chunk.
+func (r *StreamResult) Done() {
+	r.status.SetEndReason(relaycommon.StreamEndReasonDone, nil)
+	r.stopped = true
+}
+
+// IsStopped returns whether Stop() or Done() was called during this chunk.
+func (r *StreamResult) IsStopped() bool {
+	return r.stopped
+}
+
+// reset clears the per-chunk stopped flag so the object can be reused.
+func (r *StreamResult) reset() {
+	r.stopped = false
+}

relay/helper/stream_scanner.go

@@ -34,12 +34,15 @@ func getScannerBufferSize() int {
 	return DefaultMaxScannerBufferSize
 }
 
-func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo, dataHandler func(data string) bool) {
+func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo, dataHandler func(data string, sr *StreamResult)) {
 
 	if resp == nil || dataHandler == nil {
 		return
 	}
 
+	// 无条件新建 StreamStatus
+	info.StreamStatus = relaycommon.NewStreamStatus()
+
 	// 确保响应体总是被关闭
 	defer func() {
 		if resp.Body != nil {
@@ -121,6 +124,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 				wg.Done()
 				if r := recover(); r != nil {
 					logger.LogError(c, fmt.Sprintf("ping goroutine panic: %v", r))
+					info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonPanic, fmt.Errorf("ping panic: %v", r))
 					common.SafeSendBool(stopChan, true)
 				}
 				if common.DebugEnabled {
@@ -148,6 +152,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 					case err := <-done:
 						if err != nil {
 							logger.LogError(c, "ping data error: "+err.Error())
+							info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonPingFail, err)
 							return
 						}
 						if common.DebugEnabled {
@@ -155,6 +160,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 						}
 					case <-time.After(10 * time.Second):
 						logger.LogError(c, "ping data send timeout")
+						info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonPingFail, fmt.Errorf("ping send timeout"))
 						return
 					case <-ctx.Done():
 						return
@@ -184,14 +190,17 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 			wg.Done()
 			if r := recover(); r != nil {
 				logger.LogError(c, fmt.Sprintf("data handler goroutine panic: %v", r))
+				info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonPanic, fmt.Errorf("handler panic: %v", r))
 			}
 			common.SafeSendBool(stopChan, true)
 		}()
+		sr := newStreamResult(info.StreamStatus)
 		for data := range dataChan {
+			sr.reset()
 			writeMutex.Lock()
-			success := dataHandler(data)
+			dataHandler(data, sr)
 			writeMutex.Unlock()
-			if !success {
+			if sr.IsStopped() {
 				return
 			}
 		}
@@ -205,6 +214,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 			wg.Done()
 			if r := recover(); r != nil {
 				logger.LogError(c, fmt.Sprintf("scanner goroutine panic: %v", r))
+				info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonPanic, fmt.Errorf("scanner panic: %v", r))
 			}
 			common.SafeSendBool(stopChan, true)
 			if common.DebugEnabled {
@@ -220,6 +230,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 			case <-ctx.Done():
 				return
 			case <-c.Request.Context().Done():
+				info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonClientGone, c.Request.Context().Err())
 				return
 			default:
 			}
@@ -253,7 +264,7 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 					return
 				}
 			} else {
-				// done, 处理完成标志，直接退出停止读取剩余数据防止出错
+				info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonDone, nil)
 				if common.DebugEnabled {
 					println("received [DONE], stopping scanner")
 				}
@@ -264,20 +275,25 @@ func StreamScannerHandler(c *gin.Context, resp *http.Response, info *relaycommon
 		if err := scanner.Err(); err != nil {
 			if err != io.EOF {
 				logger.LogError(c, "scanner error: "+err.Error())
+				info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonScannerErr, err)
 			}
 		}
+		info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonEOF, nil)
 	})
 
 	// 主循环等待完成或超时
 	select {
 	case <-ticker.C:
-		// 超时处理逻辑
-		logger.LogError(c, "streaming timeout")
+		info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonTimeout, nil)
 	case <-stopChan:
-		// 正常结束
-		logger.LogInfo(c, "streaming finished")
+		// EndReason already set by the goroutine that triggered stopChan
 	case <-c.Request.Context().Done():
-		// 客户端断开连接
-		logger.LogInfo(c, "client disconnected")
+		info.StreamStatus.SetEndReason(relaycommon.StreamEndReasonClientGone, c.Request.Context().Err())
+	}
+
+	if info.StreamStatus.IsNormalEnd() && !info.StreamStatus.HasErrors() {
+		logger.LogInfo(c, fmt.Sprintf("stream ended: %s", info.StreamStatus.Summary()))
+	} else {
+		logger.LogError(c, fmt.Sprintf("stream ended: %s, received=%d", info.StreamStatus.Summary(), info.ReceivedResponseCount))
 	}
 }

relay/helper/stream_scanner_test.go

@@ -56,8 +56,6 @@ func buildSSEBody(n int) string {
 	return b.String()
 }
 
-// slowReader wraps a reader and injects a delay before each Read call,
-// simulating a slow upstream that trickles data.
 type slowReader struct {
 	r     io.Reader
 	delay time.Duration
@@ -79,7 +77,7 @@ func TestStreamScannerHandler_NilInputs(t *testing.T) {
 
 	info := &relaycommon.RelayInfo{ChannelMeta: &relaycommon.ChannelMeta{}}
 
-	StreamScannerHandler(c, nil, info, func(data string) bool { return true })
+	StreamScannerHandler(c, nil, info, func(data string, sr *StreamResult) {})
 	StreamScannerHandler(c, &http.Response{Body: io.NopCloser(strings.NewReader(""))}, info, nil)
 }
 
@@ -89,9 +87,8 @@ func TestStreamScannerHandler_EmptyBody(t *testing.T) {
 	c, resp, info := setupStreamTest(t, strings.NewReader(""))
 
 	var called atomic.Bool
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		called.Store(true)
-		return true
 	})
 
 	assert.False(t, called.Load(), "handler should not be called for empty body")
@@ -105,9 +102,8 @@ func TestStreamScannerHandler_1000Chunks(t *testing.T) {
 	c, resp, info := setupStreamTest(t, strings.NewReader(body))
 
 	var count atomic.Int64
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		count.Add(1)
-		return true
 	})
 
 	assert.Equal(t, int64(numChunks), count.Load())
@@ -124,9 +120,8 @@ func TestStreamScannerHandler_10000Chunks(t *testing.T) {
 	var count atomic.Int64
 	start := time.Now()
 
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		count.Add(1)
-		return true
 	})
 
 	elapsed := time.Since(start)
@@ -145,11 +140,10 @@ func TestStreamScannerHandler_OrderPreserved(t *testing.T) {
 	var mu sync.Mutex
 	received := make([]string, 0, numChunks)
 
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		mu.Lock()
 		received = append(received, data)
 		mu.Unlock()
-		return true
 	})
 
 	require.Equal(t, numChunks, len(received))
@@ -166,31 +160,32 @@ func TestStreamScannerHandler_DoneStopsScanner(t *testing.T) {
 	c, resp, info := setupStreamTest(t, strings.NewReader(body))
 
 	var count atomic.Int64
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		count.Add(1)
-		return true
 	})
 
 	assert.Equal(t, int64(50), count.Load(), "data after [DONE] must not be processed")
 }
 
-func TestStreamScannerHandler_HandlerFailureStops(t *testing.T) {
+func TestStreamScannerHandler_StopStopsStream(t *testing.T) {
 	t.Parallel()
 
 	const numChunks = 200
 	body := buildSSEBody(numChunks)
 	c, resp, info := setupStreamTest(t, strings.NewReader(body))
 
-	const failAt = 50
+	const stopAt int64 = 50
 	var count atomic.Int64
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		n := count.Add(1)
-		return n < failAt
+		if n >= stopAt {
+			sr.Stop(fmt.Errorf("fatal at %d", n))
+		}
 	})
 
-	// The worker stops at failAt; the scanner may have read ahead,
-	// but the handler should not be called beyond failAt.
-	assert.Equal(t, int64(failAt), count.Load())
+	assert.Equal(t, stopAt, count.Load())
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonHandlerStop, info.StreamStatus.EndReason)
 }
 
 func TestStreamScannerHandler_SkipsNonDataLines(t *testing.T) {
@@ -210,9 +205,8 @@ func TestStreamScannerHandler_SkipsNonDataLines(t *testing.T) {
 	c, resp, info := setupStreamTest(t, strings.NewReader(b.String()))
 
 	var count atomic.Int64
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		count.Add(1)
-		return true
 	})
 
 	assert.Equal(t, int64(100), count.Load())
@@ -225,25 +219,18 @@ func TestStreamScannerHandler_DataWithExtraSpaces(t *testing.T) {
 	c, resp, info := setupStreamTest(t, strings.NewReader(body))
 
 	var got string
-	StreamScannerHandler(c, resp, info, func(data string) bool {
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 		got = data
-		return true
 	})
 
 	assert.Equal(t, "{\"trimmed\":true}", got)
 }
 
-// ---------- Decoupling: scanner not blocked by slow handler ----------
+// ---------- Decoupling ----------
 
 func TestStreamScannerHandler_ScannerDecoupledFromSlowHandler(t *testing.T) {
 	t.Parallel()
 
-	// Strategy: use a slow upstream (io.Pipe, 10ms per chunk) AND a slow handler (20ms per chunk).
-	// If the scanner were synchronously coupled to the handler, total time would be
-	// ~numChunks * (10ms + 20ms) = 30ms * 50 = 1500ms.
-	// With decoupling, total time should be closer to
-	// ~numChunks * max(10ms, 20ms) = 20ms * 50 = 1000ms
-	// because the scanner reads ahead into the buffer while the handler processes.
 	const numChunks = 50
 	const upstreamDelay = 10 * time.Millisecond
 	const handlerDelay = 20 * time.Millisecond
@@ -273,10 +260,9 @@ func TestStreamScannerHandler_ScannerDecoupledFromSlowHandler(t *testing.T) {
 	start := time.Now()
 	done := make(chan struct{})
 	go func() {
-		StreamScannerHandler(c, resp, info, func(data string) bool {
+		StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 			time.Sleep(handlerDelay)
 			count.Add(1)
-			return true
 		})
 		close(done)
 	}()
@@ -293,7 +279,6 @@ func TestStreamScannerHandler_ScannerDecoupledFromSlowHandler(t *testing.T) {
 	coupledTime := time.Duration(numChunks) * (upstreamDelay + handlerDelay)
 	t.Logf("elapsed=%v, coupled_estimate=%v", elapsed, coupledTime)
 
-	// If decoupled, elapsed should be well under the coupled estimate.
 	assert.Less(t, elapsed, coupledTime*85/100,
 		"decoupled elapsed time (%v) should be significantly less than coupled estimate (%v)", elapsed, coupledTime)
 }
@@ -311,9 +296,8 @@ func TestStreamScannerHandler_SlowUpstreamFastHandler(t *testing.T) {
 
 	done := make(chan struct{})
 	go func() {
-		StreamScannerHandler(c, resp, info, func(data string) bool {
+		StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 			count.Add(1)
-			return true
 		})
 		close(done)
 	}()
@@ -344,8 +328,6 @@ func TestStreamScannerHandler_PingSentDuringSlowUpstream(t *testing.T) {
 		setting.PingIntervalSeconds = oldSeconds
 	})
 
-	// Create a reader that delivers data slowly: one chunk every 500ms over 3.5 seconds.
-	// The ping interval is 1s, so we should see at least 2 pings.
 	pr, pw := io.Pipe()
 	go func() {
 		defer pw.Close()
@@ -372,9 +354,8 @@ func TestStreamScannerHandler_PingSentDuringSlowUpstream(t *testing.T) {
 	var count atomic.Int64
 	done := make(chan struct{})
 	go func() {
-		StreamScannerHandler(c, resp, info, func(data string) bool {
+		StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 			count.Add(1)
-			return true
 		})
 		close(done)
 	}()
@@ -436,9 +417,8 @@ func TestStreamScannerHandler_PingDisabledByRelayInfo(t *testing.T) {
 	var count atomic.Int64
 	done := make(chan struct{})
 	go func() {
-		StreamScannerHandler(c, resp, info, func(data string) bool {
+		StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 			count.Add(1)
-			return true
 		})
 		close(done)
 	}()
@@ -456,6 +436,199 @@ func TestStreamScannerHandler_PingDisabledByRelayInfo(t *testing.T) {
 	assert.Equal(t, 0, pingCount, "pings should be disabled when DisablePing=true")
 }
 
+// ---------- StreamStatus integration ----------
+
+func TestStreamScannerHandler_StreamStatus_DoneReason(t *testing.T) {
+	t.Parallel()
+
+	body := buildSSEBody(10)
+	c, resp, info := setupStreamTest(t, strings.NewReader(body))
+
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {})
+
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonDone, info.StreamStatus.EndReason)
+	assert.Nil(t, info.StreamStatus.EndError)
+	assert.True(t, info.StreamStatus.IsNormalEnd())
+	assert.False(t, info.StreamStatus.HasErrors())
+}
+
+func TestStreamScannerHandler_StreamStatus_EOFWithoutDone(t *testing.T) {
+	t.Parallel()
+
+	var b strings.Builder
+	for i := 0; i < 5; i++ {
+		fmt.Fprintf(&b, "data: {\"id\":%d}\n", i)
+	}
+	c, resp, info := setupStreamTest(t, strings.NewReader(b.String()))
+
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {})
+
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonEOF, info.StreamStatus.EndReason)
+	assert.True(t, info.StreamStatus.IsNormalEnd())
+}
+
+func TestStreamScannerHandler_StreamStatus_HandlerStop(t *testing.T) {
+	t.Parallel()
+
+	body := buildSSEBody(100)
+	c, resp, info := setupStreamTest(t, strings.NewReader(body))
+
+	var count atomic.Int64
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
+		n := count.Add(1)
+		if n >= 10 {
+			sr.Stop(fmt.Errorf("stop at 10"))
+		}
+	})
+
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonHandlerStop, info.StreamStatus.EndReason)
+	assert.True(t, info.StreamStatus.HasErrors())
+}
+
+func TestStreamScannerHandler_StreamStatus_HandlerDone(t *testing.T) {
+	t.Parallel()
+
+	body := buildSSEBody(20)
+	c, resp, info := setupStreamTest(t, strings.NewReader(body))
+
+	var count atomic.Int64
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
+		n := count.Add(1)
+		if n >= 5 {
+			sr.Done()
+		}
+	})
+
+	assert.Equal(t, int64(5), count.Load())
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonDone, info.StreamStatus.EndReason)
+	assert.False(t, info.StreamStatus.HasErrors())
+}
+
+func TestStreamScannerHandler_StreamStatus_Timeout(t *testing.T) {
+	// Not parallel: modifies global constant.StreamingTimeout
+	oldTimeout := constant.StreamingTimeout
+	constant.StreamingTimeout = 2
+	t.Cleanup(func() { constant.StreamingTimeout = oldTimeout })
+
+	pr, pw := io.Pipe()
+	go func() {
+		fmt.Fprint(pw, "data: {\"id\":1}\n")
+		time.Sleep(10 * time.Second)
+		pw.Close()
+	}()
+
+	recorder := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(recorder)
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/chat/completions", nil)
+
+	resp := &http.Response{Body: pr}
+	info := &relaycommon.RelayInfo{ChannelMeta: &relaycommon.ChannelMeta{}}
+
+	done := make(chan struct{})
+	go func() {
+		StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {})
+		close(done)
+	}()
+
+	select {
+	case <-done:
+	case <-time.After(15 * time.Second):
+		t.Fatal("timed out waiting for stream timeout")
+	}
+
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonTimeout, info.StreamStatus.EndReason)
+	assert.False(t, info.StreamStatus.IsNormalEnd())
+}
+
+func TestStreamScannerHandler_StreamStatus_SoftErrors(t *testing.T) {
+	t.Parallel()
+
+	body := buildSSEBody(10)
+	c, resp, info := setupStreamTest(t, strings.NewReader(body))
+
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
+		sr.Error(fmt.Errorf("soft error for chunk"))
+	})
+
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonDone, info.StreamStatus.EndReason)
+	assert.True(t, info.StreamStatus.HasErrors())
+	assert.Equal(t, 10, info.StreamStatus.TotalErrorCount())
+}
+
+func TestStreamScannerHandler_StreamStatus_MultipleErrorsPerChunk(t *testing.T) {
+	t.Parallel()
+
+	body := buildSSEBody(5)
+	c, resp, info := setupStreamTest(t, strings.NewReader(body))
+
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
+		sr.Error(fmt.Errorf("error A"))
+		sr.Error(fmt.Errorf("error B"))
+	})
+
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonDone, info.StreamStatus.EndReason)
+	assert.Equal(t, 10, info.StreamStatus.TotalErrorCount())
+}
+
+func TestStreamScannerHandler_StreamStatus_ErrorThenStop(t *testing.T) {
+	t.Parallel()
+
+	// Use a large body without [DONE] to avoid race between scanner's [DONE]
+	// and handler's Stop on the sync.Once EndReason.
+	var b strings.Builder
+	for i := 0; i < 100; i++ {
+		fmt.Fprintf(&b, "data: {\"id\":%d}\n", i)
+	}
+	c, resp, info := setupStreamTest(t, strings.NewReader(b.String()))
+
+	var count atomic.Int64
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
+		count.Add(1)
+		sr.Error(fmt.Errorf("soft error"))
+		sr.Stop(fmt.Errorf("fatal"))
+	})
+
+	assert.Equal(t, int64(1), count.Load())
+	require.NotNil(t, info.StreamStatus)
+	assert.Equal(t, relaycommon.StreamEndReasonHandlerStop, info.StreamStatus.EndReason)
+	assert.Equal(t, 2, info.StreamStatus.TotalErrorCount())
+}
+
+func TestStreamScannerHandler_StreamStatus_InitializedIfNil(t *testing.T) {
+	t.Parallel()
+
+	body := buildSSEBody(1)
+	c, resp, info := setupStreamTest(t, strings.NewReader(body))
+
+	assert.Nil(t, info.StreamStatus)
+
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {})
+
+	assert.NotNil(t, info.StreamStatus)
+}
+
+func TestStreamScannerHandler_StreamStatus_PreInitialized(t *testing.T) {
+	t.Parallel()
+
+	body := buildSSEBody(5)
+	c, resp, info := setupStreamTest(t, strings.NewReader(body))
+
+	info.StreamStatus = relaycommon.NewStreamStatus()
+	info.StreamStatus.RecordError("pre-existing error")
+
+	StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {})
+
+	assert.Equal(t, relaycommon.StreamEndReasonDone, info.StreamStatus.EndReason)
+	assert.Equal(t, 1, info.StreamStatus.TotalErrorCount())
+}
+
 func TestStreamScannerHandler_PingInterleavesWithSlowUpstream(t *testing.T) {
 	t.Parallel()
 
@@ -469,9 +642,6 @@ func TestStreamScannerHandler_PingInterleavesWithSlowUpstream(t *testing.T) {
 		setting.PingIntervalSeconds = oldSeconds
 	})
 
-	// Slow upstream + slow handler. Total stream takes ~5 seconds.
-	// The ping goroutine stays alive as long as the scanner is reading,
-	// so pings should fire between data writes.
 	pr, pw := io.Pipe()
 	go func() {
 		defer pw.Close()
@@ -498,9 +668,8 @@ func TestStreamScannerHandler_PingInterleavesWithSlowUpstream(t *testing.T) {
 	var count atomic.Int64
 	done := make(chan struct{})
 	go func() {
-		StreamScannerHandler(c, resp, info, func(data string) bool {
+		StreamScannerHandler(c, resp, info, func(data string, sr *StreamResult) {
 			count.Add(1)
-			return true
 		})
 		close(done)
 	}()

relay/image_handler.go

@@ -141,6 +141,6 @@ func ImageHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *type
 		logContent = append(logContent, fmt.Sprintf("生成数量 %d", imageN))
 	}
 
-	postConsumeQuota(c, info, usage.(*dto.Usage), logContent...)
+	service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), logContent)
 	return nil
 }

relay/rerank_handler.go

@@ -96,6 +96,6 @@ func RerankHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
 		service.ResetStatusCode(newAPIError, statusCodeMappingStr)
 		return newAPIError
 	}
-	postConsumeQuota(c, info, usage.(*dto.Usage))
+	service.PostTextConsumeQuota(c, info, usage.(*dto.Usage), nil)
 	return nil
 }

relay/responses_handler.go

@@ -145,7 +145,7 @@ func ResponsesHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *
 			info.PriceData = originPriceData
 			return types.NewError(err, types.ErrorCodeModelPriceError, types.ErrOptionWithSkipRetry())
 		}
-		postConsumeQuota(c, info, usageDto)
+		service.PostTextConsumeQuota(c, info, usageDto, nil)
 
 		info.OriginModelName = originModelName
 		info.PriceData = originPriceData
@@ -155,7 +155,7 @@ func ResponsesHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *
 	if strings.HasPrefix(info.OriginModelName, "gpt-4o-audio") {
 		service.PostAudioConsumeQuota(c, info, usageDto, "")
 	} else {
-		postConsumeQuota(c, info, usageDto)
+		service.PostTextConsumeQuota(c, info, usageDto, nil)
 	}
 	return nil
 }

service/channel_affinity.go

@@ -610,14 +610,17 @@ func ShouldSkipRetryAfterChannelAffinityFailure(c *gin.Context) bool {
 		return false
 	}
 	v, ok := c.Get(ginKeyChannelAffinitySkipRetry)
+	if ok {
+		b, ok := v.(bool)
+		if ok {
+			return b
+		}
+	}
+	meta, ok := getChannelAffinityMeta(c)
 	if !ok {
 		return false
 	}
-	b, ok := v.(bool)
-	if !ok {
-		return false
-	}
-	return b
+	return meta.SkipRetry
 }
 
 func MarkChannelAffinityUsed(c *gin.Context, selectedGroup string, channelID int) {

service/channel_affinity_template_test.go

@@ -116,6 +116,66 @@ func TestApplyChannelAffinityOverrideTemplate_MergeOperations(t *testing.T) {
 	require.Equal(t, "trim_prefix", secondOp["mode"])
 }
 
+func TestShouldSkipRetryAfterChannelAffinityFailure(t *testing.T) {
+	tests := []struct {
+		name string
+		ctx  func() *gin.Context
+		want bool
+	}{
+		{
+			name: "nil context",
+			ctx: func() *gin.Context {
+				return nil
+			},
+			want: false,
+		},
+		{
+			name: "explicit skip retry flag in context",
+			ctx: func() *gin.Context {
+				ctx := buildChannelAffinityTemplateContextForTest(channelAffinityMeta{
+					RuleName:   "rule-explicit-flag",
+					SkipRetry:  false,
+					UsingGroup: "default",
+					ModelName:  "gpt-5",
+				})
+				ctx.Set(ginKeyChannelAffinitySkipRetry, true)
+				return ctx
+			},
+			want: true,
+		},
+		{
+			name: "fallback to matched rule meta",
+			ctx: func() *gin.Context {
+				return buildChannelAffinityTemplateContextForTest(channelAffinityMeta{
+					RuleName:   "rule-skip-retry",
+					SkipRetry:  true,
+					UsingGroup: "default",
+					ModelName:  "gpt-5",
+				})
+			},
+			want: true,
+		},
+		{
+			name: "no flag and no skip retry meta",
+			ctx: func() *gin.Context {
+				return buildChannelAffinityTemplateContextForTest(channelAffinityMeta{
+					RuleName:   "rule-no-skip-retry",
+					SkipRetry:  false,
+					UsingGroup: "default",
+					ModelName:  "gpt-5",
+				})
+			},
+			want: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			require.Equal(t, tt.want, ShouldSkipRetryAfterChannelAffinityFailure(tt.ctx()))
+		})
+	}
+}
+
 func TestChannelAffinityHitCodexTemplatePassHeadersEffective(t *testing.T) {
 	gin.SetMode(gin.TestMode)
 

service/convert.go

@@ -223,6 +223,25 @@ func generateStopBlock(index int) *dto.ClaudeResponse {
 	}
 }
 
+func buildClaudeUsageFromOpenAIUsage(oaiUsage *dto.Usage) *dto.ClaudeUsage {
+	if oaiUsage == nil {
+		return nil
+	}
+	usage := &dto.ClaudeUsage{
+		InputTokens:              oaiUsage.PromptTokens,
+		OutputTokens:             oaiUsage.CompletionTokens,
+		CacheCreationInputTokens: oaiUsage.PromptTokensDetails.CachedCreationTokens,
+		CacheReadInputTokens:     oaiUsage.PromptTokensDetails.CachedTokens,
+	}
+	if oaiUsage.ClaudeCacheCreation5mTokens > 0 || oaiUsage.ClaudeCacheCreation1hTokens > 0 {
+		usage.CacheCreation = &dto.ClaudeCacheCreationUsage{
+			Ephemeral5mInputTokens: oaiUsage.ClaudeCacheCreation5mTokens,
+			Ephemeral1hInputTokens: oaiUsage.ClaudeCacheCreation1hTokens,
+		}
+	}
+	return usage
+}
+
 func StreamResponseOpenAI2Claude(openAIResponse *dto.ChatCompletionsStreamResponse, info *relaycommon.RelayInfo) []*dto.ClaudeResponse {
 	if info.ClaudeConvertInfo.Done {
 		return nil
@@ -391,13 +410,8 @@ func StreamResponseOpenAI2Claude(openAIResponse *dto.ChatCompletionsStreamRespon
 			}
 			if oaiUsage != nil {
 				claudeResponses = append(claudeResponses, &dto.ClaudeResponse{
-					Type: "message_delta",
-					Usage: &dto.ClaudeUsage{
-						InputTokens:              oaiUsage.PromptTokens,
-						OutputTokens:             oaiUsage.CompletionTokens,
-						CacheCreationInputTokens: oaiUsage.PromptTokensDetails.CachedCreationTokens,
-						CacheReadInputTokens:     oaiUsage.PromptTokensDetails.CachedTokens,
-					},
+					Type:  "message_delta",
+					Usage: buildClaudeUsageFromOpenAIUsage(oaiUsage),
 					Delta: &dto.ClaudeMediaMessage{
 						StopReason: common.GetPointer[string](stopReasonOpenAI2Claude(info.FinishReason)),
 					},
@@ -419,13 +433,8 @@ func StreamResponseOpenAI2Claude(openAIResponse *dto.ChatCompletionsStreamRespon
 			oaiUsage := info.ClaudeConvertInfo.Usage
 			if oaiUsage != nil {
 				claudeResponses = append(claudeResponses, &dto.ClaudeResponse{
-					Type: "message_delta",
-					Usage: &dto.ClaudeUsage{
-						InputTokens:              oaiUsage.PromptTokens,
-						OutputTokens:             oaiUsage.CompletionTokens,
-						CacheCreationInputTokens: oaiUsage.PromptTokensDetails.CachedCreationTokens,
-						CacheReadInputTokens:     oaiUsage.PromptTokensDetails.CachedTokens,
-					},
+					Type:  "message_delta",
+					Usage: buildClaudeUsageFromOpenAIUsage(oaiUsage),
 					Delta: &dto.ClaudeMediaMessage{
 						StopReason: common.GetPointer[string](stopReasonOpenAI2Claude(info.FinishReason)),
 					},
@@ -555,13 +564,8 @@ func StreamResponseOpenAI2Claude(openAIResponse *dto.ChatCompletionsStreamRespon
 			}
 			if oaiUsage != nil {
 				claudeResponses = append(claudeResponses, &dto.ClaudeResponse{
-					Type: "message_delta",
-					Usage: &dto.ClaudeUsage{
-						InputTokens:              oaiUsage.PromptTokens,
-						OutputTokens:             oaiUsage.CompletionTokens,
-						CacheCreationInputTokens: oaiUsage.PromptTokensDetails.CachedCreationTokens,
-						CacheReadInputTokens:     oaiUsage.PromptTokensDetails.CachedTokens,
-					},
+					Type:  "message_delta",
+					Usage: buildClaudeUsageFromOpenAIUsage(oaiUsage),
 					Delta: &dto.ClaudeMediaMessage{
 						StopReason: common.GetPointer[string](stopReasonOpenAI2Claude(info.FinishReason)),
 					},
@@ -612,10 +616,7 @@ func ResponseOpenAI2Claude(openAIResponse *dto.OpenAITextResponse, info *relayco
 	}
 	claudeResponse.Content = contents
 	claudeResponse.StopReason = stopReason
-	claudeResponse.Usage = &dto.ClaudeUsage{
-		InputTokens:  openAIResponse.PromptTokens,
-		OutputTokens: openAIResponse.CompletionTokens,
-	}
+	claudeResponse.Usage = buildClaudeUsageFromOpenAIUsage(&openAIResponse.Usage)
 
 	return claudeResponse
 }

service/log_info_generate.go

@@ -73,8 +73,10 @@ func GenerateTextOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, m
 	other["admin_info"] = adminInfo
 	appendRequestPath(ctx, relayInfo, other)
 	appendRequestConversionChain(relayInfo, other)
+	appendFinalRequestFormat(relayInfo, other)
 	appendBillingInfo(relayInfo, other)
 	appendParamOverrideInfo(relayInfo, other)
+	appendStreamStatus(relayInfo, other)
 	return other
 }
 
@@ -85,6 +87,33 @@ func appendParamOverrideInfo(relayInfo *relaycommon.RelayInfo, other map[string]
 	other["po"] = relayInfo.ParamOverrideAudit
 }
 
+func appendStreamStatus(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
+	if relayInfo == nil || other == nil || !relayInfo.IsStream || relayInfo.StreamStatus == nil {
+		return
+	}
+	ss := relayInfo.StreamStatus
+	status := "ok"
+	if !ss.IsNormalEnd() || ss.HasErrors() {
+		status = "error"
+	}
+	streamInfo := map[string]interface{}{
+		"status":     status,
+		"end_reason": string(ss.EndReason),
+	}
+	if ss.EndError != nil {
+		streamInfo["end_error"] = ss.EndError.Error()
+	}
+	if ss.ErrorCount > 0 {
+		streamInfo["error_count"] = ss.ErrorCount
+		messages := make([]string, 0, len(ss.Errors))
+		for _, e := range ss.Errors {
+			messages = append(messages, e.Message)
+		}
+		streamInfo["errors"] = messages
+	}
+	other["stream_status"] = streamInfo
+}
+
 func appendBillingInfo(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
 	if relayInfo == nil || other == nil {
 		return
@@ -167,6 +196,17 @@ func appendRequestConversionChain(relayInfo *relaycommon.RelayInfo, other map[st
 	other["request_conversion"] = chain
 }
 
+func appendFinalRequestFormat(relayInfo *relaycommon.RelayInfo, other map[string]interface{}) {
+	if relayInfo == nil || other == nil {
+		return
+	}
+	if relayInfo.GetFinalRequestRelayFormat() == types.RelayFormatClaude {
+		// claude indicates the final upstream request format is Claude Messages.
+		// Frontend log rendering uses this to keep the original Claude input display.
+		other["claude"] = true
+	}
+}
+
 func GenerateWssOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.RealtimeUsage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice, userGroupRatio float64) map[string]interface{} {
 	info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, 0, 0.0, modelPrice, userGroupRatio)
 	info["ws"] = true

service/quota.go

@@ -235,108 +235,6 @@ func PostWssConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, mod
 	})
 }
 
-func PostClaudeConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.Usage) {
-	if usage != nil {
-		ObserveChannelAffinityUsageCacheByRelayFormat(ctx, usage, relayInfo.GetFinalRequestRelayFormat())
-	}
-
-	useTimeSeconds := time.Now().Unix() - relayInfo.StartTime.Unix()
-	promptTokens := usage.PromptTokens
-	completionTokens := usage.CompletionTokens
-	modelName := relayInfo.OriginModelName
-
-	tokenName := ctx.GetString("token_name")
-	completionRatio := relayInfo.PriceData.CompletionRatio
-	modelRatio := relayInfo.PriceData.ModelRatio
-	groupRatio := relayInfo.PriceData.GroupRatioInfo.GroupRatio
-	modelPrice := relayInfo.PriceData.ModelPrice
-	cacheRatio := relayInfo.PriceData.CacheRatio
-	cacheTokens := usage.PromptTokensDetails.CachedTokens
-
-	cacheCreationRatio := relayInfo.PriceData.CacheCreationRatio
-	cacheCreationRatio5m := relayInfo.PriceData.CacheCreation5mRatio
-	cacheCreationRatio1h := relayInfo.PriceData.CacheCreation1hRatio
-	cacheCreationTokens := usage.PromptTokensDetails.CachedCreationTokens
-	cacheCreationTokens5m := usage.ClaudeCacheCreation5mTokens
-	cacheCreationTokens1h := usage.ClaudeCacheCreation1hTokens
-
-	if relayInfo.ChannelType == constant.ChannelTypeOpenRouter {
-		promptTokens -= cacheTokens
-		isUsingCustomSettings := relayInfo.PriceData.UsePrice || hasCustomModelRatio(modelName, relayInfo.PriceData.ModelRatio)
-		if cacheCreationTokens == 0 && relayInfo.PriceData.CacheCreationRatio != 1 && usage.Cost != 0 && !isUsingCustomSettings {
-			maybeCacheCreationTokens := CalcOpenRouterCacheCreateTokens(*usage, relayInfo.PriceData)
-			if maybeCacheCreationTokens >= 0 && promptTokens >= maybeCacheCreationTokens {
-				cacheCreationTokens = maybeCacheCreationTokens
-			}
-		}
-		promptTokens -= cacheCreationTokens
-	}
-
-	calculateQuota := 0.0
-	if !relayInfo.PriceData.UsePrice {
-		calculateQuota = float64(promptTokens)
-		calculateQuota += float64(cacheTokens) * cacheRatio
-		calculateQuota += float64(cacheCreationTokens5m) * cacheCreationRatio5m
-		calculateQuota += float64(cacheCreationTokens1h) * cacheCreationRatio1h
-		remainingCacheCreationTokens := cacheCreationTokens - cacheCreationTokens5m - cacheCreationTokens1h
-		if remainingCacheCreationTokens > 0 {
-			calculateQuota += float64(remainingCacheCreationTokens) * cacheCreationRatio
-		}
-		calculateQuota += float64(completionTokens) * completionRatio
-		calculateQuota = calculateQuota * groupRatio * modelRatio
-	} else {
-		calculateQuota = modelPrice * common.QuotaPerUnit * groupRatio
-	}
-
-	if modelRatio != 0 && calculateQuota <= 0 {
-		calculateQuota = 1
-	}
-
-	quota := int(calculateQuota)
-
-	totalTokens := promptTokens + completionTokens
-
-	var logContent string
-	// record all the consume log even if quota is 0
-	if totalTokens == 0 {
-		// in this case, must be some error happened
-		// we cannot just return, because we may have to return the pre-consumed quota
-		quota = 0
-		logContent += fmt.Sprintf("（可能是上游出错）")
-		logger.LogError(ctx, fmt.Sprintf("total tokens is 0, cannot consume quota, userId %d, channelId %d, "+
-			"tokenId %d, model %s， pre-consumed quota %d", relayInfo.UserId, relayInfo.ChannelId, relayInfo.TokenId, modelName, relayInfo.FinalPreConsumedQuota))
-	} else {
-		model.UpdateUserUsedQuotaAndRequestCount(relayInfo.UserId, quota)
-		model.UpdateChannelUsedQuota(relayInfo.ChannelId, quota)
-	}
-
-	if err := SettleBilling(ctx, relayInfo, quota); err != nil {
-		logger.LogError(ctx, "error settling billing: "+err.Error())
-	}
-
-	other := GenerateClaudeOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio,
-		cacheTokens, cacheRatio,
-		cacheCreationTokens, cacheCreationRatio,
-		cacheCreationTokens5m, cacheCreationRatio5m,
-		cacheCreationTokens1h, cacheCreationRatio1h,
-		modelPrice, relayInfo.PriceData.GroupRatioInfo.GroupSpecialRatio)
-	model.RecordConsumeLog(ctx, relayInfo.UserId, model.RecordConsumeLogParams{
-		ChannelId:        relayInfo.ChannelId,
-		PromptTokens:     promptTokens,
-		CompletionTokens: completionTokens,
-		ModelName:        modelName,
-		TokenName:        tokenName,
-		Quota:            quota,
-		Content:          logContent,
-		TokenId:          relayInfo.TokenId,
-		UseTimeSeconds:   int(useTimeSeconds),
-		IsStream:         relayInfo.IsStream,
-		Group:            relayInfo.UsingGroup,
-		Other:            other,
-	})
-
-}
-
 func CalcOpenRouterCacheCreateTokens(usage dto.Usage, priceData types.PriceData) int {
 	if priceData.CacheCreationRatio == 1 {
 		return 0

service/text_quota.go

@@ -0,0 +1,430 @@
+package service
+
+import (
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/QuantumNous/new-api/common"
+	"github.com/QuantumNous/new-api/constant"
+	"github.com/QuantumNous/new-api/dto"
+	"github.com/QuantumNous/new-api/logger"
+	"github.com/QuantumNous/new-api/model"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	"github.com/QuantumNous/new-api/setting/operation_setting"
+	"github.com/QuantumNous/new-api/types"
+
+	"github.com/gin-gonic/gin"
+	"github.com/shopspring/decimal"
+)
+
+type textQuotaSummary struct {
+	PromptTokens             int
+	CompletionTokens         int
+	TotalTokens              int
+	CacheTokens              int
+	CacheCreationTokens      int
+	CacheCreationTokens5m    int
+	CacheCreationTokens1h    int
+	ImageTokens              int
+	AudioTokens              int
+	ModelName                string
+	TokenName                string
+	UseTimeSeconds           int64
+	CompletionRatio          float64
+	CacheRatio               float64
+	ImageRatio               float64
+	ModelRatio               float64
+	GroupRatio               float64
+	ModelPrice               float64
+	CacheCreationRatio       float64
+	CacheCreationRatio5m     float64
+	CacheCreationRatio1h     float64
+	Quota                    int
+	IsClaudeUsageSemantic    bool
+	UsageSemantic            string
+	WebSearchPrice           float64
+	WebSearchCallCount       int
+	ClaudeWebSearchPrice     float64
+	ClaudeWebSearchCallCount int
+	FileSearchPrice          float64
+	FileSearchCallCount      int
+	AudioInputPrice          float64
+	ImageGenerationCallPrice float64
+}
+
+func cacheWriteTokensTotal(summary textQuotaSummary) int {
+	if summary.CacheCreationTokens5m > 0 || summary.CacheCreationTokens1h > 0 {
+		splitCacheWriteTokens := summary.CacheCreationTokens5m + summary.CacheCreationTokens1h
+		if summary.CacheCreationTokens > splitCacheWriteTokens {
+			return summary.CacheCreationTokens
+		}
+		return splitCacheWriteTokens
+	}
+	return summary.CacheCreationTokens
+}
+
+func isLegacyClaudeDerivedOpenAIUsage(relayInfo *relaycommon.RelayInfo, usage *dto.Usage) bool {
+	if relayInfo == nil || usage == nil {
+		return false
+	}
+	if relayInfo.GetFinalRequestRelayFormat() == types.RelayFormatClaude {
+		return false
+	}
+	if usage.UsageSource != "" || usage.UsageSemantic != "" {
+		return false
+	}
+	return usage.ClaudeCacheCreation5mTokens > 0 || usage.ClaudeCacheCreation1hTokens > 0
+}
+
+func calculateTextQuotaSummary(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.Usage) textQuotaSummary {
+	summary := textQuotaSummary{
+		ModelName:            relayInfo.OriginModelName,
+		TokenName:            ctx.GetString("token_name"),
+		UseTimeSeconds:       time.Now().Unix() - relayInfo.StartTime.Unix(),
+		CompletionRatio:      relayInfo.PriceData.CompletionRatio,
+		CacheRatio:           relayInfo.PriceData.CacheRatio,
+		ImageRatio:           relayInfo.PriceData.ImageRatio,
+		ModelRatio:           relayInfo.PriceData.ModelRatio,
+		GroupRatio:           relayInfo.PriceData.GroupRatioInfo.GroupRatio,
+		ModelPrice:           relayInfo.PriceData.ModelPrice,
+		CacheCreationRatio:   relayInfo.PriceData.CacheCreationRatio,
+		CacheCreationRatio5m: relayInfo.PriceData.CacheCreation5mRatio,
+		CacheCreationRatio1h: relayInfo.PriceData.CacheCreation1hRatio,
+		UsageSemantic:        usageSemanticFromUsage(relayInfo, usage),
+	}
+	summary.IsClaudeUsageSemantic = summary.UsageSemantic == "anthropic"
+
+	if usage == nil {
+		usage = &dto.Usage{
+			PromptTokens:     relayInfo.GetEstimatePromptTokens(),
+			CompletionTokens: 0,
+			TotalTokens:      relayInfo.GetEstimatePromptTokens(),
+		}
+	}
+
+	summary.PromptTokens = usage.PromptTokens
+	summary.CompletionTokens = usage.CompletionTokens
+	summary.TotalTokens = usage.PromptTokens + usage.CompletionTokens
+	summary.CacheTokens = usage.PromptTokensDetails.CachedTokens
+	summary.CacheCreationTokens = usage.PromptTokensDetails.CachedCreationTokens
+	summary.CacheCreationTokens5m = usage.ClaudeCacheCreation5mTokens
+	summary.CacheCreationTokens1h = usage.ClaudeCacheCreation1hTokens
+	summary.ImageTokens = usage.PromptTokensDetails.ImageTokens
+	summary.AudioTokens = usage.PromptTokensDetails.AudioTokens
+	legacyClaudeDerived := isLegacyClaudeDerivedOpenAIUsage(relayInfo, usage)
+	isOpenRouterClaudeBilling := relayInfo.ChannelMeta != nil &&
+		relayInfo.ChannelType == constant.ChannelTypeOpenRouter &&
+		summary.IsClaudeUsageSemantic
+
+	if isOpenRouterClaudeBilling {
+		summary.PromptTokens -= summary.CacheTokens
+		isUsingCustomSettings := relayInfo.PriceData.UsePrice || hasCustomModelRatio(summary.ModelName, relayInfo.PriceData.ModelRatio)
+		if summary.CacheCreationTokens == 0 && relayInfo.PriceData.CacheCreationRatio != 1 && usage.Cost != 0 && !isUsingCustomSettings {
+			maybeCacheCreationTokens := CalcOpenRouterCacheCreateTokens(*usage, relayInfo.PriceData)
+			if maybeCacheCreationTokens >= 0 && summary.PromptTokens >= maybeCacheCreationTokens {
+				summary.CacheCreationTokens = maybeCacheCreationTokens
+			}
+		}
+		summary.PromptTokens -= summary.CacheCreationTokens
+	}
+
+	dPromptTokens := decimal.NewFromInt(int64(summary.PromptTokens))
+	dCacheTokens := decimal.NewFromInt(int64(summary.CacheTokens))
+	dImageTokens := decimal.NewFromInt(int64(summary.ImageTokens))
+	dAudioTokens := decimal.NewFromInt(int64(summary.AudioTokens))
+	dCompletionTokens := decimal.NewFromInt(int64(summary.CompletionTokens))
+	dCachedCreationTokens := decimal.NewFromInt(int64(summary.CacheCreationTokens))
+	dCompletionRatio := decimal.NewFromFloat(summary.CompletionRatio)
+	dCacheRatio := decimal.NewFromFloat(summary.CacheRatio)
+	dImageRatio := decimal.NewFromFloat(summary.ImageRatio)
+	dModelRatio := decimal.NewFromFloat(summary.ModelRatio)
+	dGroupRatio := decimal.NewFromFloat(summary.GroupRatio)
+	dModelPrice := decimal.NewFromFloat(summary.ModelPrice)
+	dCacheCreationRatio := decimal.NewFromFloat(summary.CacheCreationRatio)
+	dCacheCreationRatio5m := decimal.NewFromFloat(summary.CacheCreationRatio5m)
+	dCacheCreationRatio1h := decimal.NewFromFloat(summary.CacheCreationRatio1h)
+	dQuotaPerUnit := decimal.NewFromFloat(common.QuotaPerUnit)
+
+	ratio := dModelRatio.Mul(dGroupRatio)
+
+	var dWebSearchQuota decimal.Decimal
+	if relayInfo.ResponsesUsageInfo != nil {
+		if webSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists && webSearchTool.CallCount > 0 {
+			summary.WebSearchCallCount = webSearchTool.CallCount
+			summary.WebSearchPrice = operation_setting.GetWebSearchPricePerThousand(summary.ModelName, webSearchTool.SearchContextSize)
+			dWebSearchQuota = decimal.NewFromFloat(summary.WebSearchPrice).
+				Mul(decimal.NewFromInt(int64(webSearchTool.CallCount))).
+				Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
+		}
+	} else if strings.HasSuffix(summary.ModelName, "search-preview") {
+		searchContextSize := ctx.GetString("chat_completion_web_search_context_size")
+		if searchContextSize == "" {
+			searchContextSize = "medium"
+		}
+		summary.WebSearchCallCount = 1
+		summary.WebSearchPrice = operation_setting.GetWebSearchPricePerThousand(summary.ModelName, searchContextSize)
+		dWebSearchQuota = decimal.NewFromFloat(summary.WebSearchPrice).
+			Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
+	}
+
+	var dClaudeWebSearchQuota decimal.Decimal
+	summary.ClaudeWebSearchCallCount = ctx.GetInt("claude_web_search_requests")
+	if summary.ClaudeWebSearchCallCount > 0 {
+		summary.ClaudeWebSearchPrice = operation_setting.GetClaudeWebSearchPricePerThousand()
+		dClaudeWebSearchQuota = decimal.NewFromFloat(summary.ClaudeWebSearchPrice).
+			Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit).
+			Mul(decimal.NewFromInt(int64(summary.ClaudeWebSearchCallCount)))
+	}
+
+	var dFileSearchQuota decimal.Decimal
+	if relayInfo.ResponsesUsageInfo != nil {
+		if fileSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolFileSearch]; exists && fileSearchTool.CallCount > 0 {
+			summary.FileSearchCallCount = fileSearchTool.CallCount
+			summary.FileSearchPrice = operation_setting.GetFileSearchPricePerThousand()
+			dFileSearchQuota = decimal.NewFromFloat(summary.FileSearchPrice).
+				Mul(decimal.NewFromInt(int64(fileSearchTool.CallCount))).
+				Div(decimal.NewFromInt(1000)).Mul(dGroupRatio).Mul(dQuotaPerUnit)
+		}
+	}
+
+	var dImageGenerationCallQuota decimal.Decimal
+	if ctx.GetBool("image_generation_call") {
+		summary.ImageGenerationCallPrice = operation_setting.GetGPTImage1PriceOnceCall(ctx.GetString("image_generation_call_quality"), ctx.GetString("image_generation_call_size"))
+		dImageGenerationCallQuota = decimal.NewFromFloat(summary.ImageGenerationCallPrice).Mul(dGroupRatio).Mul(dQuotaPerUnit)
+	}
+
+	var audioInputQuota decimal.Decimal
+	if !relayInfo.PriceData.UsePrice {
+		baseTokens := dPromptTokens
+
+		var cachedTokensWithRatio decimal.Decimal
+		if !dCacheTokens.IsZero() {
+			if !summary.IsClaudeUsageSemantic && !legacyClaudeDerived {
+				baseTokens = baseTokens.Sub(dCacheTokens)
+			}
+			cachedTokensWithRatio = dCacheTokens.Mul(dCacheRatio)
+		}
+
+		var cachedCreationTokensWithRatio decimal.Decimal
+		hasSplitCacheCreationTokens := summary.CacheCreationTokens5m > 0 || summary.CacheCreationTokens1h > 0
+		if !dCachedCreationTokens.IsZero() || hasSplitCacheCreationTokens {
+			if !summary.IsClaudeUsageSemantic && !legacyClaudeDerived {
+				baseTokens = baseTokens.Sub(dCachedCreationTokens)
+				cachedCreationTokensWithRatio = dCachedCreationTokens.Mul(dCacheCreationRatio)
+			} else {
+				remaining := summary.CacheCreationTokens - summary.CacheCreationTokens5m - summary.CacheCreationTokens1h
+				if remaining < 0 {
+					remaining = 0
+				}
+				cachedCreationTokensWithRatio = decimal.NewFromInt(int64(remaining)).Mul(dCacheCreationRatio)
+				cachedCreationTokensWithRatio = cachedCreationTokensWithRatio.Add(decimal.NewFromInt(int64(summary.CacheCreationTokens5m)).Mul(dCacheCreationRatio5m))
+				cachedCreationTokensWithRatio = cachedCreationTokensWithRatio.Add(decimal.NewFromInt(int64(summary.CacheCreationTokens1h)).Mul(dCacheCreationRatio1h))
+			}
+		}
+
+		var imageTokensWithRatio decimal.Decimal
+		if !dImageTokens.IsZero() {
+			baseTokens = baseTokens.Sub(dImageTokens)
+			imageTokensWithRatio = dImageTokens.Mul(dImageRatio)
+		}
+
+		if !dAudioTokens.IsZero() {
+			summary.AudioInputPrice = operation_setting.GetGeminiInputAudioPricePerMillionTokens(summary.ModelName)
+			if summary.AudioInputPrice > 0 {
+				baseTokens = baseTokens.Sub(dAudioTokens)
+				audioInputQuota = decimal.NewFromFloat(summary.AudioInputPrice).
+					Div(decimal.NewFromInt(1000000)).Mul(dAudioTokens).Mul(dGroupRatio).Mul(dQuotaPerUnit)
+			}
+		}
+
+		promptQuota := baseTokens.Add(cachedTokensWithRatio).Add(imageTokensWithRatio).Add(cachedCreationTokensWithRatio)
+		completionQuota := dCompletionTokens.Mul(dCompletionRatio)
+		quotaCalculateDecimal := promptQuota.Add(completionQuota).Mul(ratio)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dWebSearchQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dClaudeWebSearchQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dFileSearchQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(audioInputQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dImageGenerationCallQuota)
+
+		if len(relayInfo.PriceData.OtherRatios) > 0 {
+			for _, otherRatio := range relayInfo.PriceData.OtherRatios {
+				quotaCalculateDecimal = quotaCalculateDecimal.Mul(decimal.NewFromFloat(otherRatio))
+			}
+		}
+
+		if !ratio.IsZero() && quotaCalculateDecimal.LessThanOrEqual(decimal.Zero) {
+			quotaCalculateDecimal = decimal.NewFromInt(1)
+		}
+		summary.Quota = int(quotaCalculateDecimal.Round(0).IntPart())
+	} else {
+		quotaCalculateDecimal := dModelPrice.Mul(dQuotaPerUnit).Mul(dGroupRatio)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dWebSearchQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dClaudeWebSearchQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dFileSearchQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(audioInputQuota)
+		quotaCalculateDecimal = quotaCalculateDecimal.Add(dImageGenerationCallQuota)
+		if len(relayInfo.PriceData.OtherRatios) > 0 {
+			for _, otherRatio := range relayInfo.PriceData.OtherRatios {
+				quotaCalculateDecimal = quotaCalculateDecimal.Mul(decimal.NewFromFloat(otherRatio))
+			}
+		}
+		summary.Quota = int(quotaCalculateDecimal.Round(0).IntPart())
+	}
+
+	if summary.TotalTokens == 0 {
+		summary.Quota = 0
+	} else if !ratio.IsZero() && summary.Quota == 0 {
+		summary.Quota = 1
+	}
+
+	return summary
+}
+
+func usageSemanticFromUsage(relayInfo *relaycommon.RelayInfo, usage *dto.Usage) string {
+	if usage != nil && usage.UsageSemantic != "" {
+		return usage.UsageSemantic
+	}
+	if relayInfo != nil && relayInfo.GetFinalRequestRelayFormat() == types.RelayFormatClaude {
+		return "anthropic"
+	}
+	return "openai"
+}
+
+func PostTextConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.Usage, extraContent []string) {
+	originUsage := usage
+	if usage == nil {
+		extraContent = append(extraContent, "上游无计费信息")
+	}
+	if originUsage != nil {
+		ObserveChannelAffinityUsageCacheByRelayFormat(ctx, usage, relayInfo.GetFinalRequestRelayFormat())
+	}
+
+	adminRejectReason := common.GetContextKeyString(ctx, constant.ContextKeyAdminRejectReason)
+	summary := calculateTextQuotaSummary(ctx, relayInfo, usage)
+
+	if summary.WebSearchCallCount > 0 {
+		extraContent = append(extraContent, fmt.Sprintf("Web Search 调用 %d 次，调用花费 %s", summary.WebSearchCallCount, decimal.NewFromFloat(summary.WebSearchPrice).Mul(decimal.NewFromInt(int64(summary.WebSearchCallCount))).Div(decimal.NewFromInt(1000)).Mul(decimal.NewFromFloat(summary.GroupRatio)).Mul(decimal.NewFromFloat(common.QuotaPerUnit)).String()))
+	}
+	if summary.ClaudeWebSearchCallCount > 0 {
+		extraContent = append(extraContent, fmt.Sprintf("Claude Web Search 调用 %d 次，调用花费 %s", summary.ClaudeWebSearchCallCount, decimal.NewFromFloat(summary.ClaudeWebSearchPrice).Div(decimal.NewFromInt(1000)).Mul(decimal.NewFromFloat(summary.GroupRatio)).Mul(decimal.NewFromFloat(common.QuotaPerUnit)).Mul(decimal.NewFromInt(int64(summary.ClaudeWebSearchCallCount))).String()))
+	}
+	if summary.FileSearchCallCount > 0 {
+		extraContent = append(extraContent, fmt.Sprintf("File Search 调用 %d 次，调用花费 %s", summary.FileSearchCallCount, decimal.NewFromFloat(summary.FileSearchPrice).Mul(decimal.NewFromInt(int64(summary.FileSearchCallCount))).Div(decimal.NewFromInt(1000)).Mul(decimal.NewFromFloat(summary.GroupRatio)).Mul(decimal.NewFromFloat(common.QuotaPerUnit)).String()))
+	}
+	if summary.AudioInputPrice > 0 && summary.AudioTokens > 0 {
+		extraContent = append(extraContent, fmt.Sprintf("Audio Input 花费 %s", decimal.NewFromFloat(summary.AudioInputPrice).Div(decimal.NewFromInt(1000000)).Mul(decimal.NewFromInt(int64(summary.AudioTokens))).Mul(decimal.NewFromFloat(summary.GroupRatio)).Mul(decimal.NewFromFloat(common.QuotaPerUnit)).String()))
+	}
+	if summary.ImageGenerationCallPrice > 0 {
+		extraContent = append(extraContent, fmt.Sprintf("Image Generation Call 花费 %s", decimal.NewFromFloat(summary.ImageGenerationCallPrice).Mul(decimal.NewFromFloat(summary.GroupRatio)).Mul(decimal.NewFromFloat(common.QuotaPerUnit)).String()))
+	}
+
+	if summary.TotalTokens == 0 {
+		extraContent = append(extraContent, "上游没有返回计费信息，无法扣费（可能是上游超时）")
+		logger.LogError(ctx, fmt.Sprintf("total tokens is 0, cannot consume quota, userId %d, channelId %d, tokenId %d, model %s， pre-consumed quota %d", relayInfo.UserId, relayInfo.ChannelId, relayInfo.TokenId, summary.ModelName, relayInfo.FinalPreConsumedQuota))
+	} else {
+		model.UpdateUserUsedQuotaAndRequestCount(relayInfo.UserId, summary.Quota)
+		model.UpdateChannelUsedQuota(relayInfo.ChannelId, summary.Quota)
+	}
+
+	if err := SettleBilling(ctx, relayInfo, summary.Quota); err != nil {
+		logger.LogError(ctx, "error settling billing: "+err.Error())
+	}
+
+	logModel := summary.ModelName
+	if strings.HasPrefix(logModel, "gpt-4-gizmo") {
+		logModel = "gpt-4-gizmo-*"
+		extraContent = append(extraContent, fmt.Sprintf("模型 %s", summary.ModelName))
+	}
+	if strings.HasPrefix(logModel, "gpt-4o-gizmo") {
+		logModel = "gpt-4o-gizmo-*"
+		extraContent = append(extraContent, fmt.Sprintf("模型 %s", summary.ModelName))
+	}
+
+	logContent := strings.Join(extraContent, ", ")
+	var other map[string]interface{}
+	if summary.IsClaudeUsageSemantic {
+		other = GenerateClaudeOtherInfo(ctx, relayInfo,
+			summary.ModelRatio, summary.GroupRatio, summary.CompletionRatio,
+			summary.CacheTokens, summary.CacheRatio,
+			summary.CacheCreationTokens, summary.CacheCreationRatio,
+			summary.CacheCreationTokens5m, summary.CacheCreationRatio5m,
+			summary.CacheCreationTokens1h, summary.CacheCreationRatio1h,
+			summary.ModelPrice, relayInfo.PriceData.GroupRatioInfo.GroupSpecialRatio)
+		other["usage_semantic"] = "anthropic"
+	} else {
+		other = GenerateTextOtherInfo(ctx, relayInfo, summary.ModelRatio, summary.GroupRatio, summary.CompletionRatio, summary.CacheTokens, summary.CacheRatio, summary.ModelPrice, relayInfo.PriceData.GroupRatioInfo.GroupSpecialRatio)
+	}
+	if adminRejectReason != "" {
+		other["reject_reason"] = adminRejectReason
+	}
+	if summary.ImageTokens != 0 {
+		other["image"] = true
+		other["image_ratio"] = summary.ImageRatio
+		other["image_output"] = summary.ImageTokens
+	}
+	if summary.WebSearchCallCount > 0 {
+		other["web_search"] = true
+		other["web_search_call_count"] = summary.WebSearchCallCount
+		other["web_search_price"] = summary.WebSearchPrice
+	} else if summary.ClaudeWebSearchCallCount > 0 {
+		other["web_search"] = true
+		other["web_search_call_count"] = summary.ClaudeWebSearchCallCount
+		other["web_search_price"] = summary.ClaudeWebSearchPrice
+	}
+	if summary.FileSearchCallCount > 0 {
+		other["file_search"] = true
+		other["file_search_call_count"] = summary.FileSearchCallCount
+		other["file_search_price"] = summary.FileSearchPrice
+	}
+	if summary.AudioInputPrice > 0 && summary.AudioTokens > 0 {
+		other["audio_input_seperate_price"] = true
+		other["audio_input_token_count"] = summary.AudioTokens
+		other["audio_input_price"] = summary.AudioInputPrice
+	}
+	if summary.ImageGenerationCallPrice > 0 {
+		other["image_generation_call"] = true
+		other["image_generation_call_price"] = summary.ImageGenerationCallPrice
+	}
+	if summary.CacheCreationTokens > 0 {
+		other["cache_creation_tokens"] = summary.CacheCreationTokens
+		other["cache_creation_ratio"] = summary.CacheCreationRatio
+	}
+	if summary.CacheCreationTokens5m > 0 {
+		other["cache_creation_tokens_5m"] = summary.CacheCreationTokens5m
+		other["cache_creation_ratio_5m"] = summary.CacheCreationRatio5m
+	}
+	if summary.CacheCreationTokens1h > 0 {
+		other["cache_creation_tokens_1h"] = summary.CacheCreationTokens1h
+		other["cache_creation_ratio_1h"] = summary.CacheCreationRatio1h
+	}
+	cacheWriteTokens := cacheWriteTokensTotal(summary)
+	if cacheWriteTokens > 0 {
+		// cache_write_tokens: normalized cache creation total for UI display.
+		// If split 5m/1h values are present, this is their sum; otherwise it falls back
+		// to cache_creation_tokens.
+		other["cache_write_tokens"] = cacheWriteTokens
+	}
+	if relayInfo.GetFinalRequestRelayFormat() != types.RelayFormatClaude && usage != nil && usage.UsageSource != "" && usage.InputTokens > 0 {
+		// input_tokens_total: explicit normalized total input used by the usage log UI.
+		// Only write this field when upstream/current conversion has already provided a
+		// reliable total input value and tagged the usage source. Do not infer it from
+		// prompt/cache fields here, otherwise old upstream payloads may be double-counted.
+		other["input_tokens_total"] = usage.InputTokens
+	}
+
+	model.RecordConsumeLog(ctx, relayInfo.UserId, model.RecordConsumeLogParams{
+		ChannelId:        relayInfo.ChannelId,
+		PromptTokens:     summary.PromptTokens,
+		CompletionTokens: summary.CompletionTokens,
+		ModelName:        logModel,
+		TokenName:        summary.TokenName,
+		Quota:            summary.Quota,
+		Content:          logContent,
+		TokenId:          relayInfo.TokenId,
+		UseTimeSeconds:   int(summary.UseTimeSeconds),
+		IsStream:         relayInfo.IsStream,
+		Group:            relayInfo.UsingGroup,
+		Other:            other,
+	})
+}

service/text_quota_test.go

@@ -0,0 +1,318 @@
+package service
+
+import (
+	"net/http/httptest"
+	"testing"
+	"time"
+
+	"github.com/QuantumNous/new-api/constant"
+	"github.com/QuantumNous/new-api/dto"
+	relaycommon "github.com/QuantumNous/new-api/relay/common"
+	"github.com/QuantumNous/new-api/types"
+
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+)
+
+func TestCalculateTextQuotaSummaryUnifiedForClaudeSemantic(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(w)
+
+	usage := &dto.Usage{
+		PromptTokens:     1000,
+		CompletionTokens: 200,
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedTokens:         100,
+			CachedCreationTokens: 50,
+		},
+		ClaudeCacheCreation5mTokens: 10,
+		ClaudeCacheCreation1hTokens: 20,
+	}
+
+	priceData := types.PriceData{
+		ModelRatio:           1,
+		CompletionRatio:      2,
+		CacheRatio:           0.1,
+		CacheCreationRatio:   1.25,
+		CacheCreation5mRatio: 1.25,
+		CacheCreation1hRatio: 2,
+		GroupRatioInfo: types.GroupRatioInfo{
+			GroupRatio: 1,
+		},
+	}
+
+	chatRelayInfo := &relaycommon.RelayInfo{
+		RelayFormat:             types.RelayFormatOpenAI,
+		FinalRequestRelayFormat: types.RelayFormatClaude,
+		OriginModelName:         "claude-3-7-sonnet",
+		PriceData:               priceData,
+		StartTime:               time.Now(),
+	}
+	messageRelayInfo := &relaycommon.RelayInfo{
+		RelayFormat:             types.RelayFormatClaude,
+		FinalRequestRelayFormat: types.RelayFormatClaude,
+		OriginModelName:         "claude-3-7-sonnet",
+		PriceData:               priceData,
+		StartTime:               time.Now(),
+	}
+
+	chatSummary := calculateTextQuotaSummary(ctx, chatRelayInfo, usage)
+	messageSummary := calculateTextQuotaSummary(ctx, messageRelayInfo, usage)
+
+	require.Equal(t, messageSummary.Quota, chatSummary.Quota)
+	require.Equal(t, messageSummary.CacheCreationTokens5m, chatSummary.CacheCreationTokens5m)
+	require.Equal(t, messageSummary.CacheCreationTokens1h, chatSummary.CacheCreationTokens1h)
+	require.True(t, chatSummary.IsClaudeUsageSemantic)
+	require.Equal(t, 1488, chatSummary.Quota)
+}
+
+func TestCalculateTextQuotaSummaryUsesSplitClaudeCacheCreationRatios(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(w)
+
+	relayInfo := &relaycommon.RelayInfo{
+		RelayFormat:             types.RelayFormatOpenAI,
+		FinalRequestRelayFormat: types.RelayFormatClaude,
+		OriginModelName:         "claude-3-7-sonnet",
+		PriceData: types.PriceData{
+			ModelRatio:           1,
+			CompletionRatio:      1,
+			CacheRatio:           0,
+			CacheCreationRatio:   1,
+			CacheCreation5mRatio: 2,
+			CacheCreation1hRatio: 3,
+			GroupRatioInfo: types.GroupRatioInfo{
+				GroupRatio: 1,
+			},
+		},
+		StartTime: time.Now(),
+	}
+
+	usage := &dto.Usage{
+		PromptTokens:     100,
+		CompletionTokens: 0,
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedCreationTokens: 10,
+		},
+		ClaudeCacheCreation5mTokens: 2,
+		ClaudeCacheCreation1hTokens: 3,
+	}
+
+	summary := calculateTextQuotaSummary(ctx, relayInfo, usage)
+
+	// 100 + remaining(5)*1 + 2*2 + 3*3 = 118
+	require.Equal(t, 118, summary.Quota)
+}
+
+func TestCalculateTextQuotaSummaryUsesAnthropicUsageSemanticFromUpstreamUsage(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(w)
+
+	relayInfo := &relaycommon.RelayInfo{
+		RelayFormat:     types.RelayFormatOpenAI,
+		OriginModelName: "claude-3-7-sonnet",
+		PriceData: types.PriceData{
+			ModelRatio:           1,
+			CompletionRatio:      2,
+			CacheRatio:           0.1,
+			CacheCreationRatio:   1.25,
+			CacheCreation5mRatio: 1.25,
+			CacheCreation1hRatio: 2,
+			GroupRatioInfo: types.GroupRatioInfo{
+				GroupRatio: 1,
+			},
+		},
+		StartTime: time.Now(),
+	}
+
+	usage := &dto.Usage{
+		PromptTokens:     1000,
+		CompletionTokens: 200,
+		UsageSemantic:    "anthropic",
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedTokens:         100,
+			CachedCreationTokens: 50,
+		},
+		ClaudeCacheCreation5mTokens: 10,
+		ClaudeCacheCreation1hTokens: 20,
+	}
+
+	summary := calculateTextQuotaSummary(ctx, relayInfo, usage)
+
+	require.True(t, summary.IsClaudeUsageSemantic)
+	require.Equal(t, "anthropic", summary.UsageSemantic)
+	require.Equal(t, 1488, summary.Quota)
+}
+
+func TestCacheWriteTokensTotal(t *testing.T) {
+	t.Run("split cache creation", func(t *testing.T) {
+		summary := textQuotaSummary{
+			CacheCreationTokens:   50,
+			CacheCreationTokens5m: 10,
+			CacheCreationTokens1h: 20,
+		}
+		require.Equal(t, 50, cacheWriteTokensTotal(summary))
+	})
+
+	t.Run("legacy cache creation", func(t *testing.T) {
+		summary := textQuotaSummary{CacheCreationTokens: 50}
+		require.Equal(t, 50, cacheWriteTokensTotal(summary))
+	})
+
+	t.Run("split cache creation without aggregate remainder", func(t *testing.T) {
+		summary := textQuotaSummary{
+			CacheCreationTokens5m: 10,
+			CacheCreationTokens1h: 20,
+		}
+		require.Equal(t, 30, cacheWriteTokensTotal(summary))
+	})
+}
+
+func TestCalculateTextQuotaSummaryHandlesLegacyClaudeDerivedOpenAIUsage(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(w)
+
+	relayInfo := &relaycommon.RelayInfo{
+		RelayFormat:     types.RelayFormatOpenAI,
+		OriginModelName: "claude-3-7-sonnet",
+		PriceData: types.PriceData{
+			ModelRatio:           1,
+			CompletionRatio:      5,
+			CacheRatio:           0.1,
+			CacheCreationRatio:   1.25,
+			CacheCreation5mRatio: 1.25,
+			CacheCreation1hRatio: 2,
+			GroupRatioInfo:       types.GroupRatioInfo{GroupRatio: 1},
+		},
+		StartTime: time.Now(),
+	}
+
+	usage := &dto.Usage{
+		PromptTokens:     62,
+		CompletionTokens: 95,
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedTokens: 3544,
+		},
+		ClaudeCacheCreation5mTokens: 586,
+	}
+
+	summary := calculateTextQuotaSummary(ctx, relayInfo, usage)
+
+	// 62 + 3544*0.1 + 586*1.25 + 95*5 = 1624.9 => 1624
+	require.Equal(t, 1624, summary.Quota)
+}
+
+func TestCalculateTextQuotaSummarySeparatesOpenRouterCacheReadFromPromptBilling(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(w)
+
+	relayInfo := &relaycommon.RelayInfo{
+		OriginModelName: "openai/gpt-4.1",
+		ChannelMeta: &relaycommon.ChannelMeta{
+			ChannelType: constant.ChannelTypeOpenRouter,
+		},
+		PriceData: types.PriceData{
+			ModelRatio:         1,
+			CompletionRatio:    1,
+			CacheRatio:         0.1,
+			CacheCreationRatio: 1.25,
+			GroupRatioInfo:     types.GroupRatioInfo{GroupRatio: 1},
+		},
+		StartTime: time.Now(),
+	}
+
+	usage := &dto.Usage{
+		PromptTokens:     2604,
+		CompletionTokens: 383,
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedTokens: 2432,
+		},
+	}
+
+	summary := calculateTextQuotaSummary(ctx, relayInfo, usage)
+
+	// OpenRouter OpenAI-format display keeps prompt_tokens as total input,
+	// but billing still separates normal input from cache read tokens.
+	// quota = (2604 - 2432) + 2432*0.1 + 383 = 798.2 => 798
+	require.Equal(t, 2604, summary.PromptTokens)
+	require.Equal(t, 798, summary.Quota)
+}
+
+func TestCalculateTextQuotaSummarySeparatesOpenRouterCacheCreationFromPromptBilling(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(w)
+
+	relayInfo := &relaycommon.RelayInfo{
+		OriginModelName: "openai/gpt-4.1",
+		ChannelMeta: &relaycommon.ChannelMeta{
+			ChannelType: constant.ChannelTypeOpenRouter,
+		},
+		PriceData: types.PriceData{
+			ModelRatio:         1,
+			CompletionRatio:    1,
+			CacheCreationRatio: 1.25,
+			GroupRatioInfo:     types.GroupRatioInfo{GroupRatio: 1},
+		},
+		StartTime: time.Now(),
+	}
+
+	usage := &dto.Usage{
+		PromptTokens:     2604,
+		CompletionTokens: 383,
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedCreationTokens: 100,
+		},
+	}
+
+	summary := calculateTextQuotaSummary(ctx, relayInfo, usage)
+
+	// prompt_tokens is still logged as total input, but cache creation is billed separately.
+	// quota = (2604 - 100) + 100*1.25 + 383 = 3012
+	require.Equal(t, 2604, summary.PromptTokens)
+	require.Equal(t, 3012, summary.Quota)
+}
+
+func TestCalculateTextQuotaSummaryKeepsPrePRClaudeOpenRouterBilling(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	ctx, _ := gin.CreateTestContext(w)
+
+	relayInfo := &relaycommon.RelayInfo{
+		FinalRequestRelayFormat: types.RelayFormatClaude,
+		OriginModelName:         "anthropic/claude-3.7-sonnet",
+		ChannelMeta: &relaycommon.ChannelMeta{
+			ChannelType: constant.ChannelTypeOpenRouter,
+		},
+		PriceData: types.PriceData{
+			ModelRatio:         1,
+			CompletionRatio:    1,
+			CacheRatio:         0.1,
+			CacheCreationRatio: 1.25,
+			GroupRatioInfo:     types.GroupRatioInfo{GroupRatio: 1},
+		},
+		StartTime: time.Now(),
+	}
+
+	usage := &dto.Usage{
+		PromptTokens:     2604,
+		CompletionTokens: 383,
+		PromptTokensDetails: dto.InputTokenDetails{
+			CachedTokens: 2432,
+		},
+	}
+
+	summary := calculateTextQuotaSummary(ctx, relayInfo, usage)
+
+	// Pre-PR PostClaudeConsumeQuota behavior for OpenRouter:
+	// prompt = 2604 - 2432 = 172
+	// quota = 172 + 2432*0.1 + 383 = 798.2 => 798
+	require.True(t, summary.IsClaudeUsageSemantic)
+	require.Equal(t, 172, summary.PromptTokens)
+	require.Equal(t, 798, summary.Quota)
+}

setting/operation_setting/channel_affinity_setting.go

@@ -88,7 +88,7 @@ var channelAffinitySetting = ChannelAffinitySetting{
 			ValueRegex:            "",
 			TTLSeconds:            0,
 			ParamOverrideTemplate: buildPassHeaderTemplate(codexCliPassThroughHeaders),
-			SkipRetryOnFailure:    false,
+			SkipRetryOnFailure:    true,
 			IncludeUsingGroup:     true,
 			IncludeRuleName:       true,
 			UserAgentInclude:      nil,
@@ -103,7 +103,7 @@ var channelAffinitySetting = ChannelAffinitySetting{
 			ValueRegex:            "",
 			TTLSeconds:            0,
 			ParamOverrideTemplate: buildPassHeaderTemplate(claudeCliPassThroughHeaders),
-			SkipRetryOnFailure:    false,
+			SkipRetryOnFailure:    true,
 			IncludeUsingGroup:     true,
 			IncludeRuleName:       true,
 			UserAgentInclude:      nil,

setting/performance_setting/config.go

@@ -36,7 +36,7 @@ var performanceSetting = PerformanceSetting{
 	MonitorEnabled:         true,
 	MonitorCPUThreshold:    90,
 	MonitorMemoryThreshold: 90,
-	MonitorDiskThreshold:   90,
+	MonitorDiskThreshold:   95,
 }
 
 func init() {

web/src/components/auth/OAuth2Callback.jsx

@@ -56,7 +56,7 @@ const OAuth2Callback = (props) => {
         return;
       }
 
-      if (message === 'bind') {
+      if (data?.action === 'bind') {
         showSuccess(t('绑定成功！'));
         navigate('/console/personal');
       } else {

web/src/components/table/channels/ChannelsColumnDefs.jsx

@@ -538,19 +538,24 @@ export const getChannelsColumns = ({
                 </Tooltip>
                 <Tooltip
                   content={
-                    t('剩余额度') +
-                    ': ' +
-                    renderQuotaWithAmount(record.balance) +
-                    t('，点击更新')
+                    record.type === 57
+                      ? t('查看 Codex 帐号信息与用量')
+                      : t('剩余额度') +
+                        ': ' +
+                        renderQuotaWithAmount(record.balance) +
+                        t('，点击更新')
                   }
                 >
                   <Tag
-                    color='white'
-                    type='ghost'
+                    color={record.type === 57 ? 'light-blue' : 'white'}
+                    type={record.type === 57 ? 'light' : 'ghost'}
                     shape='circle'
+                    className={record.type === 57 ? 'cursor-pointer' : ''}
                     onClick={() => updateChannelBalance(record)}
                   >
-                    {renderQuotaWithAmount(record.balance)}
+                    {record.type === 57
+                      ? t('帐号信息')
+                      : renderQuotaWithAmount(record.balance)}
                   </Tag>
                 </Tooltip>
               </Space>

web/src/components/table/channels/modals/CodexUsageModal.jsx

@@ -22,9 +22,11 @@ import {
   Modal,
   Button,
   Progress,
-  Tag,
   Typography,
   Spin,
+  Tag,
+  Descriptions,
+  Collapse,
 } from '@douyinfe/semi-ui';
 import { API, showError } from '../../../../helpers';
 
@@ -128,6 +130,87 @@ const formatUnixSeconds = (unixSeconds) => {
   }
 };
 
+const getDisplayText = (value) => {
+  if (value == null) return '';
+  return String(value).trim();
+};
+
+const formatAccountTypeLabel = (value, t) => {
+  const tt = typeof t === 'function' ? t : (v) => v;
+  const normalized = normalizePlanType(value);
+  switch (normalized) {
+    case 'free':
+      return 'Free';
+    case 'plus':
+      return 'Plus';
+    case 'pro':
+      return 'Pro';
+    case 'team':
+      return 'Team';
+    case 'enterprise':
+      return 'Enterprise';
+    default:
+      return getDisplayText(value) || tt('未识别');
+  }
+};
+
+const getAccountTypeTagColor = (value) => {
+  const normalized = normalizePlanType(value);
+  switch (normalized) {
+    case 'enterprise':
+      return 'green';
+    case 'team':
+      return 'cyan';
+    case 'pro':
+      return 'blue';
+    case 'plus':
+      return 'violet';
+    case 'free':
+      return 'amber';
+    default:
+      return 'grey';
+  }
+};
+
+const resolveUsageStatusTag = (t, rateLimit) => {
+  const tt = typeof t === 'function' ? t : (v) => v;
+  if (!rateLimit || Object.keys(rateLimit).length === 0) {
+    return <Tag color='grey'>{tt('待确认')}</Tag>;
+  }
+  if (rateLimit?.allowed && !rateLimit?.limit_reached) {
+    return <Tag color='green'>{tt('可用')}</Tag>;
+  }
+  return <Tag color='red'>{tt('受限')}</Tag>;
+};
+
+const AccountInfoValue = ({ t, value, onCopy, monospace = false }) => {
+  const tt = typeof t === 'function' ? t : (v) => v;
+  const text = getDisplayText(value);
+  const hasValue = text !== '';
+
+  return (
+    <div className='flex min-w-0 items-start justify-between gap-2'>
+      <div
+        className={`min-w-0 flex-1 break-all text-xs leading-5 text-semi-color-text-1 ${
+          monospace ? 'font-mono' : ''
+        }`}
+      >
+        {hasValue ? text : '-'}
+      </div>
+      <Button
+        size='small'
+        type='tertiary'
+        theme='borderless'
+        className='shrink-0 px-1 text-xs'
+        disabled={!hasValue}
+        onClick={() => onCopy?.(text)}
+      >
+        {tt('复制')}
+      </Button>
+    </div>
+  );
+};
+
 const RateLimitWindowCard = ({ t, title, windowData }) => {
   const tt = typeof t === 'function' ? t : (v) => v;
   const hasWindowData =
@@ -181,50 +264,100 @@ const RateLimitWindowCard = ({ t, title, windowData }) => {
 
 const CodexUsageView = ({ t, record, payload, onCopy, onRefresh }) => {
   const tt = typeof t === 'function' ? t : (v) => v;
+  const [showRawJson, setShowRawJson] = useState(false);
   const data = payload?.data ?? null;
   const rateLimit = data?.rate_limit ?? {};
   const { fiveHourWindow, weeklyWindow } = resolveRateLimitWindows(data);
-
-  const allowed = !!rateLimit?.allowed;
-  const limitReached = !!rateLimit?.limit_reached;
   const upstreamStatus = payload?.upstream_status;
-
-  const statusTag =
-    allowed && !limitReached ? (
-      <Tag color='green'>{tt('可用')}</Tag>
-    ) : (
-      <Tag color='red'>{tt('受限')}</Tag>
-    );
+  const accountType = data?.plan_type ?? rateLimit?.plan_type;
+  const accountTypeLabel = formatAccountTypeLabel(accountType, tt);
+  const accountTypeTagColor = getAccountTypeTagColor(accountType);
+  const statusTag = resolveUsageStatusTag(tt, rateLimit);
+  const userId = data?.user_id;
+  const email = data?.email;
+  const accountId = data?.account_id;
+  const errorMessage =
+    payload?.success === false ? getDisplayText(payload?.message) || tt('获取用量失败') : '';
 
   const rawText =
     typeof data === 'string' ? data : JSON.stringify(data ?? payload, null, 2);
 
   return (
-    <div className='flex flex-col gap-3'>
-      <div className='flex flex-wrap items-center justify-between gap-2'>
-        <Text type='tertiary' size='small'>
-          {tt('渠道：')}
-          {record?.name || '-'} ({tt('编号：')}
-          {record?.id || '-'})
-        </Text>
-        <div className='flex items-center gap-2'>
-          {statusTag}
-          <Button
-            size='small'
-            type='tertiary'
-            theme='borderless'
-            onClick={onRefresh}
-          >
+    <div className='flex flex-col gap-4'>
+      {errorMessage && (
+        <div className='rounded-xl border border-red-200 bg-red-50 px-4 py-3 text-sm text-red-700'>
+          {errorMessage}
+        </div>
+      )}
+
+      <div className='rounded-xl border border-semi-color-border bg-semi-color-bg-0 p-3'>
+        <div className='flex flex-wrap items-start justify-between gap-2'>
+          <div className='min-w-0'>
+            <div className='text-xs font-medium text-semi-color-text-2'>
+              {tt('Codex 帐号')}
+            </div>
+            <div className='mt-2 flex flex-wrap items-center gap-2'>
+              <Tag
+                color={accountTypeTagColor}
+                type='light'
+                shape='circle'
+                size='large'
+                className='font-semibold'
+              >
+                {accountTypeLabel}
+              </Tag>
+              {statusTag}
+              <Tag color='grey' type='light' shape='circle'>
+                {tt('上游状态码：')}
+                {upstreamStatus ?? '-'}
+              </Tag>
+            </div>
+          </div>
+          <Button size='small' type='tertiary' theme='outline' onClick={onRefresh}>
             {tt('刷新')}
           </Button>
         </div>
+
+        <div className='mt-2 rounded-lg bg-semi-color-fill-0 px-3 py-2'>
+          <Descriptions>
+            <Descriptions.Item itemKey='User ID'>
+              <AccountInfoValue
+                t={tt}
+                value={userId}
+                onCopy={onCopy}
+                monospace={true}
+              />
+            </Descriptions.Item>
+            <Descriptions.Item itemKey={tt('邮箱')}>
+              <AccountInfoValue t={tt} value={email} onCopy={onCopy} />
+            </Descriptions.Item>
+            <Descriptions.Item itemKey='Account ID'>
+              <AccountInfoValue
+                t={tt}
+                value={accountId}
+                onCopy={onCopy}
+                monospace={true}
+              />
+            </Descriptions.Item>
+          </Descriptions>
+        </div>
+
+        <div className='mt-2 text-xs text-semi-color-text-2'>
+          {tt('渠道：')}
+          {record?.name || '-'} ({tt('编号：')}
+          {record?.id || '-'})
+        </div>
       </div>
 
-      <div className='flex flex-wrap items-center justify-between gap-2'>
-        <Text type='tertiary' size='small'>
-          {tt('上游状态码：')}
-          {upstreamStatus ?? '-'}
-        </Text>
+      <div>
+        <div className='mb-2'>
+          <div className='text-sm font-semibold text-semi-color-text-0'>
+            {tt('额度窗口')}
+          </div>
+          <Text type='tertiary' size='small'>
+            {tt('用于观察当前帐号在 Codex 上游的限额使用情况')}
+          </Text>
+        </div>
       </div>
 
       <div className='grid grid-cols-1 gap-3 md:grid-cols-2'>
@@ -240,23 +373,30 @@ const CodexUsageView = ({ t, record, payload, onCopy, onRefresh }) => {
         />
       </div>
 
-      <div>
-        <div className='mb-1 flex items-center justify-between gap-2'>
-          <div className='text-sm font-medium'>{tt('原始 JSON')}</div>
-          <Button
-            size='small'
-            type='primary'
-            theme='outline'
-            onClick={() => onCopy?.(rawText)}
-            disabled={!rawText}
-          >
-            {tt('复制')}
-          </Button>
-        </div>
-        <pre className='max-h-[50vh] overflow-auto rounded-lg bg-semi-color-fill-0 p-3 text-xs text-semi-color-text-0'>
-          {rawText}
-        </pre>
-      </div>
+      <Collapse
+        activeKey={showRawJson ? ['raw-json'] : []}
+        onChange={(activeKey) => {
+          const keys = Array.isArray(activeKey) ? activeKey : [activeKey];
+          setShowRawJson(keys.includes('raw-json'));
+        }}
+      >
+        <Collapse.Panel header={tt('原始 JSON')} itemKey='raw-json'>
+          <div className='mb-2 flex justify-end'>
+            <Button
+              size='small'
+              type='primary'
+              theme='outline'
+              onClick={() => onCopy?.(rawText)}
+              disabled={!rawText}
+            >
+              {tt('复制')}
+            </Button>
+          </div>
+          <pre className='max-h-[50vh] overflow-y-auto rounded-lg bg-semi-color-fill-0 p-3 text-xs text-semi-color-text-0'>
+            {rawText}
+          </pre>
+        </Collapse.Panel>
+      </Collapse>
     </div>
   );
 };
@@ -351,7 +491,7 @@ export const openCodexUsageModal = ({ t, record, payload, onCopy }) => {
   const tt = typeof t === 'function' ? t : (v) => v;
 
   Modal.info({
-    title: tt('Codex 用量'),
+    title: tt('Codex 帐号与用量'),
     centered: true,
     width: 900,
     style: { maxWidth: '95vw' },

web/src/constants/channel-affinity-template.constants.js

@@ -68,7 +68,7 @@ export const CHANNEL_AFFINITY_RULE_TEMPLATES = {
     param_override_template: CODEX_CLI_HEADER_PASSTHROUGH_TEMPLATE,
     value_regex: '',
     ttl_seconds: 0,
-    skip_retry_on_failure: false,
+    skip_retry_on_failure: true,
     include_using_group: true,
     include_rule_name: true,
   },
@@ -80,7 +80,7 @@ export const CHANNEL_AFFINITY_RULE_TEMPLATES = {
     param_override_template: CLAUDE_CLI_HEADER_PASSTHROUGH_TEMPLATE,
     value_regex: '',
     ttl_seconds: 0,
-    skip_retry_on_failure: false,
+    skip_retry_on_failure: true,
     include_using_group: true,
     include_rule_name: true,
   },

web/src/hooks/usage-logs/useUsageLogsData.jsx

@@ -601,6 +601,32 @@ export const useLogsData = () => {
           value: other.request_path,
         });
       }
+      if (isAdminUser && other?.stream_status) {
+        const ss = other.stream_status;
+        const isOk = ss.status === 'ok';
+        const statusLabel = isOk ? '✓ ' + t('正常') : '✗ ' + t('异常');
+        let streamValue = statusLabel + ' (' + (ss.end_reason || 'unknown') + ')';
+        if (ss.error_count > 0) {
+          streamValue += ` [${t('软错误')}: ${ss.error_count}]`;
+        }
+        if (ss.end_error) {
+          streamValue += ` - ${ss.end_error}`;
+        }
+        expandDataLocal.push({
+          key: t('流状态'),
+          value: streamValue,
+        });
+        if (Array.isArray(ss.errors) && ss.errors.length > 0) {
+          expandDataLocal.push({
+            key: t('流错误详情'),
+            value: (
+              <div style={{ maxWidth: 600, whiteSpace: 'pre-line', wordBreak: 'break-word', lineHeight: 1.6 }}>
+                {ss.errors.join('\n')}
+              </div>
+            ),
+          });
+        }
+      }
       if (Array.isArray(other?.po) && other.po.length > 0) {
         expandDataLocal.push({
           key: t('参数覆盖'),

web/src/i18n/i18n.js

@@ -52,4 +52,6 @@ i18n
     },
   });
 
+window.__i18n = i18n;
+
 export default i18n;

web/src/i18n/locales/en.json

@@ -1507,6 +1507,8 @@
     "收益统计": "Income statistics",
     "收起": "Collapse",
     "收起侧边栏": "Collapse sidebar",
+    "向左展开": "Expand left",
+    "向右展开": "Expand right",
     "收起内容": "Collapse content",
     "放大": "Upscalers",
     "放大编辑": "Expand editor",
@@ -2676,6 +2678,11 @@
     "请求结束后多退少补": "Adjust after request completion",
     "请求超时，请刷新页面后重新发起 GitHub 登录": "Request timed out, please refresh and restart GitHub login",
     "请求路径": "Request path",
+    "流状态": "Stream Status",
+    "流错误详情": "Stream Error Details",
+    "软错误": "soft errors",
+    "正常": "Normal",
+    "异常": "Abnormal",
     "请求转换": "Request conversion",
     "请求预扣费额度": "Pre-deduction quota for requests",
     "请点击我": "Please click me",
@@ -3220,6 +3227,12 @@
     "高级设置": "Advanced Settings",
     "高级选项": "Advanced Options",
     "高级配置": "Advanced Configuration",
+    "核心配置": "Core Configuration",
+    "创建渠道所需的基本信息": "Basic information needed to create a channel",
+    "请求配置": "Request Configuration",
+    "渠道行为": "Channel Behavior",
+    "额外设置": "Extra Settings",
+    "上游模型管理": "Upstream Model Management",
     "黑名单": "Blacklist",
     "默认": "Default",
     "默认 API 版本": "Default API Version",
@@ -3337,6 +3350,8 @@
     "输入价格：{{symbol}}{{price}} / 1M tokens": "Input Price: {{symbol}}{{price}} / 1M tokens",
     "输出价格 {{symbol}}{{price}} / 1M tokens": "Output Price {{symbol}}{{price}} / 1M tokens",
     "输出价格：{{symbol}}{{price}} / 1M tokens": "Output Price: {{symbol}}{{price}} / 1M tokens",
-    "输出价格：{{symbol}}{{total}} / 1M tokens": "Output Price: {{symbol}}{{total}} / 1M tokens"
+    "输出价格：{{symbol}}{{total}} / 1M tokens": "Output Price: {{symbol}}{{total}} / 1M tokens",
+    "例如：gpt-4.1-nano,regex:^claude-.*$,regex:^sora-.*$": "Example: gpt-4.1-nano,regex:^claude-.*$,regex:^sora-.*$",
+    "支持精确匹配；使用 regex: 开头可按正则匹配。": "Supports exact matching. Use a regex: prefix for regex matching."
   }
 }

web/src/i18n/locales/fr.json

@@ -1495,6 +1495,8 @@
     "收益统计": "Statistiques sur les revenus",
     "收起": "Réduire",
     "收起侧边栏": "Réduire la barre latérale",
+    "向左展开": "Développer à gauche",
+    "向右展开": "Développer à droite",
     "收起内容": "Réduire le contenu",
     "放大": "Upscalers",
     "放大编辑": "Développer l'éditeur",
@@ -3164,6 +3166,12 @@
     "高级设置": "Paramètres avancés",
     "高级选项": "Options avancées",
     "高级配置": "Advanced Configuration",
+    "核心配置": "Configuration principale",
+    "创建渠道所需的基本信息": "Informations de base pour créer un canal",
+    "请求配置": "Configuration des requêtes",
+    "渠道行为": "Comportement du canal",
+    "额外设置": "Paramètres supplémentaires",
+    "上游模型管理": "Gestion des modèles amont",
     "黑名单": "Liste noire",
     "默认": "Par défaut",
     "默认 API 版本": "Version de l'API par défaut",

web/src/i18n/locales/ja.json

@@ -1478,6 +1478,8 @@
     "收益统计": "収益統計",
     "收起": "折りたたみ",
     "收起侧边栏": "サイドバー折りたたみ",
+    "向左展开": "左に展開",
+    "向右展开": "右に展開",
     "收起内容": "コンテンツ折りたたみ",
     "放大": "アップスケール",
     "放大编辑": "エディタで開く",
@@ -3145,6 +3147,12 @@
     "高级设置": "詳細設定",
     "高级选项": "高度なオプション",
     "高级配置": "Advanced Configuration",
+    "核心配置": "基本設定",
+    "创建渠道所需的基本信息": "チャネル作成に必要な基本情報",
+    "请求配置": "リクエスト設定",
+    "渠道行为": "チャネル動作",
+    "额外设置": "追加設定",
+    "上游模型管理": "上流モデル管理",
     "黑名单": "ブラックリスト",
     "默认": "デフォルト",
     "默认 API 版本": "デフォルトAPIバージョン",

web/src/i18n/locales/ru.json

@@ -1507,6 +1507,8 @@
     "收益统计": "Статистика доходов",
     "收起": "Свернуть",
     "收起侧边栏": "Свернуть боковую панель",
+    "向左展开": "Развернуть влево",
+    "向右展开": "Развернуть вправо",
     "收起内容": "Свернуть содержимое",
     "放大": "Увеличить",
     "放大编辑": "Увеличить и редактировать",
@@ -3178,6 +3180,12 @@
     "高级设置": "Расширенные настройки",
     "高级选项": "Расширенные параметры",
     "高级配置": "Advanced Configuration",
+    "核心配置": "Основные настройки",
+    "创建渠道所需的基本信息": "Основная информация для создания канала",
+    "请求配置": "Настройки запросов",
+    "渠道行为": "Поведение канала",
+    "额外设置": "Дополнительные настройки",
+    "上游模型管理": "Управление моделями апстрима",
     "黑名单": "Черный список",
     "默认": "По умолчанию",
     "默认 API 版本": "Версия API по умолчанию",

web/src/i18n/locales/vi.json

@@ -1479,6 +1479,8 @@
     "收益统计": "Thống kê thu nhập",
     "收起": "Thu gọn",
     "收起侧边栏": "Thu gọn thanh bên",
+    "向左展开": "Mở rộng sang trái",
+    "向右展开": "Mở rộng sang phải",
     "收起内容": "Thu gọn nội dung",
     "放大": "Upscalers",
     "放大编辑": "Mở rộng trình chỉnh sửa",
@@ -3716,6 +3718,12 @@
     "高级设置": "Cài đặt nâng cao",
     "高级选项": "Tùy chọn nâng cao",
     "高级配置": "Advanced Configuration",
+    "核心配置": "Cấu hình cốt lõi",
+    "创建渠道所需的基本信息": "Thông tin cơ bản cần thiết để tạo kênh",
+    "请求配置": "Cấu hình yêu cầu",
+    "渠道行为": "Hành vi kênh",
+    "额外设置": "Cài đặt bổ sung",
+    "上游模型管理": "Quản lý mô hình thượng nguồn",
     "黑名单": "Danh sách đen",
     "默认": "Mặc định",
     "默认 API 版本": "Phiên bản API mặc định",

web/src/i18n/locales/zh-TW.json

@@ -1190,6 +1190,8 @@
     "收益统计": "收益統計",
     "收起": "收起",
     "收起侧边栏": "收起側邊欄",
+    "向左展开": "向左展開",
+    "向右展开": "向右展開",
     "收起内容": "收起內容",
     "放大": "放大",
     "放大编辑": "放大編輯",
@@ -2653,6 +2655,12 @@
     "验证配置错误": "驗證設定錯誤",
     "高级设置": "進階設定",
     "高级配置": "進階設定",
+    "核心配置": "核心設定",
+    "创建渠道所需的基本信息": "建立頻道所需的基本資訊",
+    "请求配置": "請求設定",
+    "渠道行为": "頻道行為",
+    "额外设置": "額外設定",
+    "上游模型管理": "上游模型管理",
     "黑名单": "黑名單",
     "默认": "預設",
     "默认 API 版本": "預設 API 版本",

web/src/index.css

@@ -971,3 +971,26 @@ html.dark .with-pastel-balls::before {
 .semi-datepicker-range-input {
   border-radius: 10px !important;
 }
+
+@keyframes slideInLeft {
+  from {
+    transform: translateX(-100%);
+    opacity: 0;
+  }
+  to {
+    transform: translateX(0);
+    opacity: 1;
+  }
+}
+
+/* Channel advanced settings side panel animations */
+@keyframes slideInRight {
+  from {
+    transform: translateX(100%);
+    opacity: 0;
+  }
+  to {
+    transform: translateX(0);
+    opacity: 1;
+  }
+}

web/src/pages/Setting/Operation/SettingsChannelAffinity.jsx

@@ -539,6 +539,15 @@ export default function SettingsChannelAffinity(props) {
       dataIndex: 'ttl_seconds',
       render: (v) => <Text>{Number(v || 0) || '-'}</Text>,
     },
+    {
+      title: t('失败后不重试'),
+      dataIndex: 'skip_retry_on_failure',
+      render: (value) => (
+        <Tag color={value ? 'orange' : 'grey'} style={{ marginRight: 4 }}>
+          {value ? t('是') : t('否')}
+        </Tag>
+      ),
+    },
     {
       title: t('覆盖模板'),
       render: (_, record) => {
@@ -1096,6 +1105,18 @@ export default function SettingsChannelAffinity(props) {
             </Col>
           </Row>
 
+          <Row gutter={16} style={{ marginTop: 12 }}>
+            <Col xs={24} sm={12}>
+              <Form.Switch
+                field='skip_retry_on_failure'
+                label={t('失败后不重试')}
+              />
+              <Text type='tertiary' size='small'>
+                {t('开启后，若该规则命中且请求失败，将不会切换渠道重试。')}
+              </Text>
+            </Col>
+          </Row>
+
           <Collapse
             keepDOM
             activeKey={modalAdvancedActiveKey}
@@ -1251,18 +1272,6 @@ export default function SettingsChannelAffinity(props) {
                   </Text>
                 </Col>
               </Row>
-
-              <Row gutter={16}>
-                <Col xs={24} sm={12}>
-                  <Form.Switch
-                    field='skip_retry_on_failure'
-                    label={t('失败后不重试')}
-                  />
-                  <Text type='tertiary' size='small'>
-                    {t('开启后，若该规则命中且请求失败，将不会切换渠道重试。')}
-                  </Text>
-                </Col>
-              </Row>
             </Collapse.Panel>
           </Collapse>
 

web/src/pages/Setting/Performance/SettingsPerformance.jsx

@@ -71,7 +71,7 @@ export default function SettingsPerformance(props) {
     'performance_setting.monitor_enabled': false,
     'performance_setting.monitor_cpu_threshold': 90,
     'performance_setting.monitor_memory_threshold': 90,
-    'performance_setting.monitor_disk_threshold': 90,
+    'performance_setting.monitor_disk_threshold': 95,
   });
   const refForm = useRef();
   const [inputsRow, setInputsRow] = useState(inputs);