Merge pull request #1522 from QuantumNous/support-deepseek-claude

feat: support deepseek claude format (convert)

constant/context_key.go

@@ -11,7 +11,6 @@ const (
 	ContextKeyTokenKey               ContextKey = "token_key"
 	ContextKeyTokenId                ContextKey = "token_id"
 	ContextKeyTokenGroup             ContextKey = "token_group"
-	ContextKeyTokenAllowIps          ContextKey = "allow_ips"
 	ContextKeyTokenSpecificChannelId ContextKey = "specific_channel_id"
 	ContextKeyTokenModelLimitEnabled ContextKey = "token_model_limit_enabled"
 	ContextKeyTokenModelLimit        ContextKey = "token_model_limit"

controller/channel-test.go

@@ -161,7 +161,7 @@ func testChannel(channel *model.Channel, testModel string) testResult {
 	logInfo.ApiKey = ""
 	common.SysLog(fmt.Sprintf("testing channel %d with model %s , info %+v ", channel.Id, testModel, logInfo))
 
-	priceData, err := helper.ModelPriceHelper(c, info, 0, int(request.MaxTokens))
+	priceData, err := helper.ModelPriceHelper(c, info, 0, int(request.GetMaxTokens()))
 	if err != nil {
 		return testResult{
 			context:     c,
@@ -275,7 +275,7 @@ func testChannel(channel *model.Channel, testModel string) testResult {
 		Quota:            quota,
 		Content:          "模型测试",
 		UseTimeSeconds:   int(consumedTime),
-		IsStream:         false,
+		IsStream:         info.IsStream,
 		Group:            info.UsingGroup,
 		Other:            other,
 	})

controller/channel.go

@@ -1107,6 +1107,10 @@ func ManageMultiKeys(c *gin.Context) {
 		return
 	}
 
+	lock := model.GetChannelPollingLock(channel.Id)
+	lock.Lock()
+	defer lock.Unlock()
+
 	switch request.Action {
 	case "get_key_status":
 		keys := channel.GetKeys()

dto/claude.go

@@ -361,7 +361,7 @@ type ClaudeUsage struct {
 	CacheCreationInputTokens int                  `json:"cache_creation_input_tokens"`
 	CacheReadInputTokens     int                  `json:"cache_read_input_tokens"`
 	OutputTokens             int                  `json:"output_tokens"`
-	ServerToolUse            *ClaudeServerToolUse `json:"server_tool_use"`
+	ServerToolUse            *ClaudeServerToolUse `json:"server_tool_use,omitempty"`
 }
 
 type ClaudeServerToolUse struct {

dto/openai_request.go

@@ -99,8 +99,11 @@ type StreamOptions struct {
 	IncludeUsage bool `json:"include_usage,omitempty"`
 }
 
-func (r *GeneralOpenAIRequest) GetMaxTokens() int {
-	return int(r.MaxTokens)
+func (r *GeneralOpenAIRequest) GetMaxTokens() uint {
+	if r.MaxCompletionTokens != 0 {
+		return r.MaxCompletionTokens
+	}
+	return r.MaxTokens
 }
 
 func (r *GeneralOpenAIRequest) ParseInput() []string {

go.mod

@@ -7,9 +7,10 @@ require (
 	github.com/Calcium-Ion/go-epay v0.0.4
 	github.com/andybalholm/brotli v1.1.1
 	github.com/anknown/ahocorasick v0.0.0-20190904063843-d75dbd5169c0
-	github.com/aws/aws-sdk-go-v2 v1.26.1
+	github.com/aws/aws-sdk-go-v2 v1.37.2
 	github.com/aws/aws-sdk-go-v2/credentials v1.17.11
-	github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4
+	github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.33.0
+	github.com/aws/smithy-go v1.22.5
 	github.com/bytedance/gopkg v0.0.0-20220118071334-3db87571198b
 	github.com/gin-contrib/cors v1.7.2
 	github.com/gin-contrib/gzip v0.0.6
@@ -24,6 +25,7 @@ require (
 	github.com/gorilla/websocket v1.5.0
 	github.com/joho/godotenv v1.5.1
 	github.com/pkg/errors v0.9.1
+	github.com/pquerna/otp v1.5.0
 	github.com/samber/lo v1.39.0
 	github.com/shirou/gopsutil v3.21.11+incompatible
 	github.com/shopspring/decimal v1.4.0
@@ -41,10 +43,9 @@ require (
 
 require (
 	github.com/anknown/darts v0.0.0-20151216065714-83ff685239e6 // indirect
-	github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.6.2 // indirect
-	github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.5 // indirect
-	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.5 // indirect
-	github.com/aws/smithy-go v1.20.2 // indirect
+	github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.0 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.2 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.2 // indirect
 	github.com/boombuler/barcode v1.1.0 // indirect
 	github.com/bytedance/sonic v1.11.6 // indirect
 	github.com/bytedance/sonic/loader v0.1.1 // indirect
@@ -80,7 +81,6 @@ require (
 	github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd // indirect
 	github.com/modern-go/reflect2 v1.0.2 // indirect
 	github.com/pelletier/go-toml/v2 v2.2.1 // indirect
-	github.com/pquerna/otp v1.5.0 // indirect
 	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
 	github.com/tklauser/go-sysconf v0.3.12 // indirect
 	github.com/tklauser/numcpus v0.6.1 // indirect

go.sum

@@ -6,21 +6,20 @@ github.com/anknown/ahocorasick v0.0.0-20190904063843-d75dbd5169c0 h1:onfun1RA+Kc
 github.com/anknown/ahocorasick v0.0.0-20190904063843-d75dbd5169c0/go.mod h1:4yg+jNTYlDEzBjhGS96v+zjyA3lfXlFd5CiTLIkPBLI=
 github.com/anknown/darts v0.0.0-20151216065714-83ff685239e6 h1:HblK3eJHq54yET63qPCTJnks3loDse5xRmmqHgHzwoI=
 github.com/anknown/darts v0.0.0-20151216065714-83ff685239e6/go.mod h1:pbiaLIeYLUbgMY1kwEAdwO6UKD5ZNwdPGQlwokS9fe8=
-github.com/aws/aws-sdk-go-v2 v1.26.1 h1:5554eUqIYVWpU0YmeeYZ0wU64H2VLBs8TlhRB2L+EkA=
-github.com/aws/aws-sdk-go-v2 v1.26.1/go.mod h1:ffIFB97e2yNsv4aTSGkqtHnppsIJzw7G7BReUZ3jCXM=
-github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.6.2 h1:x6xsQXGSmW6frevwDA+vi/wqhp1ct18mVXYN08/93to=
-github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.6.2/go.mod h1:lPprDr1e6cJdyYeGXnRaJoP4Md+cDBvi2eOj00BlGmg=
+github.com/aws/aws-sdk-go-v2 v1.37.2 h1:xkW1iMYawzcmYFYEV0UCMxc8gSsjCGEhBXQkdQywVbo=
+github.com/aws/aws-sdk-go-v2 v1.37.2/go.mod h1:9Q0OoGQoboYIAJyslFyF1f5K1Ryddop8gqMhWx/n4Wg=
+github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.0 h1:6GMWV6CNpA/6fbFHnoAjrv4+LGfyTqZz2LtCHnspgDg=
+github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.0/go.mod h1:/mXlTIVG9jbxkqDnr5UQNQxW1HRYxeGklkM9vAFeabg=
 github.com/aws/aws-sdk-go-v2/credentials v1.17.11 h1:YuIB1dJNf1Re822rriUOTxopaHHvIq0l/pX3fwO+Tzs=
 github.com/aws/aws-sdk-go-v2/credentials v1.17.11/go.mod h1:AQtFPsDH9bI2O+71anW6EKL+NcD7LG3dpKGMV4SShgo=
-github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.5 h1:aw39xVGeRWlWx9EzGVnhOR4yOjQDHPQ6o6NmBlscyQg=
-github.com/aws/aws-sdk-go-v2/internal/configsources v1.3.5/go.mod h1:FSaRudD0dXiMPK2UjknVwwTYyZMRsHv3TtkabsZih5I=
-github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.5 h1:PG1F3OD1szkuQPzDw3CIQsRIrtTlUC3lP84taWzHlq0=
-github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.6.5/go.mod h1:jU1li6RFryMz+so64PpKtudI+QzbKoIEivqdf6LNpOc=
-github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4 h1:JgHnonzbnA3pbqj76wYsSZIZZQYBxkmMEjvL6GHy8XU=
-github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.7.4/go.mod h1:nZspkhg+9p8iApLFoyAqfyuMP0F38acy2Hm3r5r95Cg=
-github.com/aws/smithy-go v1.20.2 h1:tbp628ireGtzcHDDmLT/6ADHidqnwgF57XOXZe6tp4Q=
-github.com/aws/smithy-go v1.20.2/go.mod h1:krry+ya/rV9RDcV/Q16kpu6ypI4K2czasz0NC3qS14E=
-github.com/boombuler/barcode v1.0.1-0.20190219062509-6c824513bacc h1:biVzkmvwrH8WK8raXaxBx6fRVTlJILwEwQGL1I/ByEI=
+github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.2 h1:sPiRHLVUIIQcoVZTNwqQcdtjkqkPopyYmIX0M5ElRf4=
+github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.2/go.mod h1:ik86P3sgV+Bk7c1tBFCwI3VxMoSEwl4YkRB9xn1s340=
+github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.2 h1:ZdzDAg075H6stMZtbD2o+PyB933M/f20e9WmCBC17wA=
+github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.2/go.mod h1:eE1IIzXG9sdZCB0pNNpMpsYTLl4YdOQD3njiVN1e/E4=
+github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.33.0 h1:JzidOz4Hcn2RbP5fvIS1iAP+DcRv5VJtgixbEYDsI5g=
+github.com/aws/aws-sdk-go-v2/service/bedrockruntime v1.33.0/go.mod h1:9A4/PJYlWjvjEzzoOLGQjkLt4bYK9fRWi7uz1GSsAcA=
+github.com/aws/smithy-go v1.22.5 h1:P9ATCXPMb2mPjYBgueqJNCA5S9UfktsW0tTxi+a7eqw=
+github.com/aws/smithy-go v1.22.5/go.mod h1:t1ufH5HMublsJYulve2RKmHDC15xu1f26kHCp/HgceI=
 github.com/boombuler/barcode v1.0.1-0.20190219062509-6c824513bacc/go.mod h1:paBWMcWSl3LHKBqUq+rly7CNSldXjb2rDl3JlRe0mD8=
 github.com/boombuler/barcode v1.1.0 h1:ChaYjBR63fr4LFyGn8E8nt7dBSt3MiU3zMOZqFvVkHo=
 github.com/boombuler/barcode v1.1.0/go.mod h1:paBWMcWSl3LHKBqUq+rly7CNSldXjb2rDl3JlRe0mD8=

middleware/auth.go

@@ -4,7 +4,10 @@ import (
 	"fmt"
 	"net/http"
 	"one-api/common"
+	"one-api/constant"
 	"one-api/model"
+	"one-api/setting"
+	"one-api/setting/ratio_setting"
 	"strconv"
 	"strings"
 
@@ -234,6 +237,16 @@ func TokenAuth() func(c *gin.Context) {
 			abortWithOpenAiMessage(c, http.StatusUnauthorized, err.Error())
 			return
 		}
+
+		allowIpsMap := token.GetIpLimitsMap()
+		if len(allowIpsMap) != 0 {
+			clientIp := c.ClientIP()
+			if _, ok := allowIpsMap[clientIp]; !ok {
+				abortWithOpenAiMessage(c, http.StatusForbidden, "您的 IP 不在令牌允许访问的列表中")
+				return
+			}
+		}
+
 		userCache, err := model.GetUserCache(token.UserId)
 		if err != nil {
 			abortWithOpenAiMessage(c, http.StatusInternalServerError, err.Error())
@@ -247,6 +260,25 @@ func TokenAuth() func(c *gin.Context) {
 
 		userCache.WriteContext(c)
 
+		userGroup := userCache.Group
+		tokenGroup := token.Group
+		if tokenGroup != "" {
+			// check common.UserUsableGroups[userGroup]
+			if _, ok := setting.GetUserUsableGroups(userGroup)[tokenGroup]; !ok {
+				abortWithOpenAiMessage(c, http.StatusForbidden, fmt.Sprintf("令牌分组 %s 已被禁用", tokenGroup))
+				return
+			}
+			// check group in common.GroupRatio
+			if !ratio_setting.ContainsGroupRatio(tokenGroup) {
+				if tokenGroup != "auto" {
+					abortWithOpenAiMessage(c, http.StatusForbidden, fmt.Sprintf("分组 %s 已被弃用", tokenGroup))
+					return
+				}
+			}
+			userGroup = tokenGroup
+		}
+		common.SetContextKey(c, constant.ContextKeyUsingGroup, userGroup)
+
 		err = SetupContextForToken(c, token, parts...)
 		if err != nil {
 			return
@@ -273,7 +305,6 @@ func SetupContextForToken(c *gin.Context, token *model.Token, parts ...string) e
 	} else {
 		c.Set("token_model_limit_enabled", false)
 	}
-	c.Set("allow_ips", token.GetIpLimitsMap())
 	c.Set("token_group", token.Group)
 	if len(parts) > 1 {
 		if model.IsAdmin(token.UserId) {

middleware/distributor.go

@@ -10,7 +10,6 @@ import (
 	"one-api/model"
 	relayconstant "one-api/relay/constant"
 	"one-api/service"
-	"one-api/setting"
 	"one-api/setting/ratio_setting"
 	"one-api/types"
 	"strconv"
@@ -27,14 +26,6 @@ type ModelRequest struct {
 
 func Distribute() func(c *gin.Context) {
 	return func(c *gin.Context) {
-		allowIpsMap := common.GetContextKeyStringMap(c, constant.ContextKeyTokenAllowIps)
-		if len(allowIpsMap) != 0 {
-			clientIp := c.ClientIP()
-			if _, ok := allowIpsMap[clientIp]; !ok {
-				abortWithOpenAiMessage(c, http.StatusForbidden, "您的 IP 不在令牌允许访问的列表中")
-				return
-			}
-		}
 		var channel *model.Channel
 		channelId, ok := common.GetContextKey(c, constant.ContextKeyTokenSpecificChannelId)
 		modelRequest, shouldSelectChannel, err := getModelRequest(c)
@@ -42,24 +33,6 @@ func Distribute() func(c *gin.Context) {
 			abortWithOpenAiMessage(c, http.StatusBadRequest, "Invalid request, "+err.Error())
 			return
 		}
-		userGroup := common.GetContextKeyString(c, constant.ContextKeyUserGroup)
-		tokenGroup := common.GetContextKeyString(c, constant.ContextKeyTokenGroup)
-		if tokenGroup != "" {
-			// check common.UserUsableGroups[userGroup]
-			if _, ok := setting.GetUserUsableGroups(userGroup)[tokenGroup]; !ok {
-				abortWithOpenAiMessage(c, http.StatusForbidden, fmt.Sprintf("令牌分组 %s 已被禁用", tokenGroup))
-				return
-			}
-			// check group in common.GroupRatio
-			if !ratio_setting.ContainsGroupRatio(tokenGroup) {
-				if tokenGroup != "auto" {
-					abortWithOpenAiMessage(c, http.StatusForbidden, fmt.Sprintf("分组 %s 已被弃用", tokenGroup))
-					return
-				}
-			}
-			userGroup = tokenGroup
-		}
-		common.SetContextKey(c, constant.ContextKeyUsingGroup, userGroup)
 		if ok {
 			id, err := strconv.Atoi(channelId.(string))
 			if err != nil {
@@ -81,22 +54,21 @@ func Distribute() func(c *gin.Context) {
 			modelLimitEnable := common.GetContextKeyBool(c, constant.ContextKeyTokenModelLimitEnabled)
 			if modelLimitEnable {
 				s, ok := common.GetContextKey(c, constant.ContextKeyTokenModelLimit)
-				var tokenModelLimit map[string]bool
-				if ok {
-					tokenModelLimit = s.(map[string]bool)
-				} else {
-					tokenModelLimit = map[string]bool{}
-				}
-				if tokenModelLimit != nil {
-					if _, ok := tokenModelLimit[modelRequest.Model]; !ok {
-						abortWithOpenAiMessage(c, http.StatusForbidden, "该令牌无权访问模型 "+modelRequest.Model)
-						return
-					}
-				} else {
+				if !ok {
 					// token model limit is empty, all models are not allowed
 					abortWithOpenAiMessage(c, http.StatusForbidden, "该令牌无权访问任何模型")
 					return
 				}
+				var tokenModelLimit map[string]bool
+				tokenModelLimit, ok = s.(map[string]bool)
+				if !ok {
+					tokenModelLimit = map[string]bool{}
+				}
+				matchName := ratio_setting.FormatMatchingModelName(modelRequest.Model) // match gpts & thinking-*
+				if _, ok := tokenModelLimit[matchName]; !ok {
+					abortWithOpenAiMessage(c, http.StatusForbidden, "该令牌无权访问模型 "+modelRequest.Model)
+					return
+				}
 			}
 
 			if shouldSelectChannel {
@@ -105,6 +77,7 @@ func Distribute() func(c *gin.Context) {
 					return
 				}
 				var selectGroup string
+				userGroup := common.GetContextKeyString(c, constant.ContextKeyUsingGroup)
 				channel, selectGroup, err = model.CacheGetRandomSatisfiedChannel(c, userGroup, modelRequest.Model, 0)
 				if err != nil {
 					showGroup := userGroup

model/channel.go

@@ -141,7 +141,7 @@ func (channel *Channel) GetNextEnabledKey() (string, int, *types.NewAPIError) {
 		return keys[selectedIdx], selectedIdx, nil
 	case constant.MultiKeyModePolling:
 		// Use channel-specific lock to ensure thread-safe polling
-		lock := getChannelPollingLock(channel.Id)
+		lock := GetChannelPollingLock(channel.Id)
 		lock.Lock()
 		defer lock.Unlock()
 
@@ -500,8 +500,8 @@ var channelStatusLock sync.Mutex
 // channelPollingLocks stores locks for each channel.id to ensure thread-safe polling
 var channelPollingLocks sync.Map
 
-// getChannelPollingLock returns or creates a mutex for the given channel ID
-func getChannelPollingLock(channelId int) *sync.Mutex {
+// GetChannelPollingLock returns or creates a mutex for the given channel ID
+func GetChannelPollingLock(channelId int) *sync.Mutex {
 	if lock, exists := channelPollingLocks.Load(channelId); exists {
 		return lock.(*sync.Mutex)
 	}

model/channel_cache.go

@@ -7,6 +7,7 @@ import (
 	"one-api/common"
 	"one-api/constant"
 	"one-api/setting"
+	"one-api/setting/ratio_setting"
 	"sort"
 	"strings"
 	"sync"
@@ -128,12 +129,7 @@ func CacheGetRandomSatisfiedChannel(c *gin.Context, group string, model string,
 }
 
 func getRandomSatisfiedChannel(group string, model string, retry int) (*Channel, error) {
-	if strings.HasPrefix(model, "gpt-4-gizmo") {
-		model = "gpt-4-gizmo-*"
-	}
-	if strings.HasPrefix(model, "gpt-4o-gizmo") {
-		model = "gpt-4o-gizmo-*"
-	}
+	model = ratio_setting.FormatMatchingModelName(model)
 
 	// if memory cache is disabled, get channel directly from database
 	if !common.MemoryCacheEnabled {

relay/channel/ali/adaptor.go

@@ -3,16 +3,17 @@ package ali
 import (
 	"errors"
 	"fmt"
+	"github.com/gin-gonic/gin"
 	"io"
 	"net/http"
 	"one-api/dto"
 	"one-api/relay/channel"
+	"one-api/relay/channel/claude"
 	"one-api/relay/channel/openai"
 	relaycommon "one-api/relay/common"
 	"one-api/relay/constant"
 	"one-api/types"
-
-	"github.com/gin-gonic/gin"
+	"strings"
 )
 
 type Adaptor struct {
@@ -23,10 +24,8 @@ func (a *Adaptor) ConvertGeminiRequest(*gin.Context, *relaycommon.RelayInfo, *dt
 	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) ConvertClaudeRequest(*gin.Context, *relaycommon.RelayInfo, *dto.ClaudeRequest) (any, error) {
-	//TODO implement me
-	panic("implement me")
-	return nil, nil
+func (a *Adaptor) ConvertClaudeRequest(c *gin.Context, info *relaycommon.RelayInfo, req *dto.ClaudeRequest) (any, error) {
+	return req, nil
 }
 
 func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
@@ -34,18 +33,24 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	var fullRequestURL string
-	switch info.RelayMode {
-	case constant.RelayModeEmbeddings:
-		fullRequestURL = fmt.Sprintf("%s/compatible-mode/v1/embeddings", info.BaseUrl)
-	case constant.RelayModeRerank:
-		fullRequestURL = fmt.Sprintf("%s/api/v1/services/rerank/text-rerank/text-rerank", info.BaseUrl)
-	case constant.RelayModeImagesGenerations:
-		fullRequestURL = fmt.Sprintf("%s/api/v1/services/aigc/text2image/image-synthesis", info.BaseUrl)
-	case constant.RelayModeCompletions:
-		fullRequestURL = fmt.Sprintf("%s/compatible-mode/v1/completions", info.BaseUrl)
+	switch info.RelayFormat {
+	case relaycommon.RelayFormatClaude:
+		fullRequestURL = fmt.Sprintf("%s/api/v2/apps/claude-code-proxy/v1/messages", info.BaseUrl)
 	default:
-		fullRequestURL = fmt.Sprintf("%s/compatible-mode/v1/chat/completions", info.BaseUrl)
+		switch info.RelayMode {
+		case constant.RelayModeEmbeddings:
+			fullRequestURL = fmt.Sprintf("%s/compatible-mode/v1/embeddings", info.BaseUrl)
+		case constant.RelayModeRerank:
+			fullRequestURL = fmt.Sprintf("%s/api/v1/services/rerank/text-rerank/text-rerank", info.BaseUrl)
+		case constant.RelayModeImagesGenerations:
+			fullRequestURL = fmt.Sprintf("%s/api/v1/services/aigc/text2image/image-synthesis", info.BaseUrl)
+		case constant.RelayModeCompletions:
+			fullRequestURL = fmt.Sprintf("%s/compatible-mode/v1/completions", info.BaseUrl)
+		default:
+			fullRequestURL = fmt.Sprintf("%s/compatible-mode/v1/chat/completions", info.BaseUrl)
+		}
 	}
+
 	return fullRequestURL, nil
 }
 
@@ -65,7 +70,13 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 	if request == nil {
 		return nil, errors.New("request is nil")
 	}
-
+	// docs: https://bailian.console.aliyun.com/?tab=api#/api/?type=model&url=2712216
+	// fix: InternalError.Algo.InvalidParameter: The value of the enable_thinking parameter is restricted to True.
+	if strings.Contains(request.Model, "thinking") {
+		request.EnableThinking = true
+		request.Stream = true
+		info.IsStream = true
+	}
 	// fix: ali parameter.enable_thinking must be set to false for non-streaming calls
 	if !info.IsStream {
 		request.EnableThinking = false
@@ -106,18 +117,27 @@ func (a *Adaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, request
 }
 
 func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (usage any, err *types.NewAPIError) {
-	switch info.RelayMode {
-	case constant.RelayModeImagesGenerations:
-		err, usage = aliImageHandler(c, resp, info)
-	case constant.RelayModeEmbeddings:
-		err, usage = aliEmbeddingHandler(c, resp)
-	case constant.RelayModeRerank:
-		err, usage = RerankHandler(c, resp, info)
-	default:
+	switch info.RelayFormat {
+	case relaycommon.RelayFormatClaude:
 		if info.IsStream {
-			usage, err = openai.OaiStreamHandler(c, info, resp)
+			err, usage = claude.ClaudeStreamHandler(c, resp, info, claude.RequestModeMessage)
 		} else {
-			usage, err = openai.OpenaiHandler(c, info, resp)
+			err, usage = claude.ClaudeHandler(c, resp, info, claude.RequestModeMessage)
+		}
+	default:
+		switch info.RelayMode {
+		case constant.RelayModeImagesGenerations:
+			err, usage = aliImageHandler(c, resp, info)
+		case constant.RelayModeEmbeddings:
+			err, usage = aliEmbeddingHandler(c, resp)
+		case constant.RelayModeRerank:
+			err, usage = RerankHandler(c, resp, info)
+		default:
+			if info.IsStream {
+				usage, err = openai.OaiStreamHandler(c, info, resp)
+			} else {
+				usage, err = openai.OpenaiHandler(c, info, resp)
+			}
 		}
 	}
 	return

relay/channel/aws/constants.go

@@ -13,6 +13,7 @@ var awsModelIDMap = map[string]string{
 	"claude-3-7-sonnet-20250219": "anthropic.claude-3-7-sonnet-20250219-v1:0",
 	"claude-sonnet-4-20250514":   "anthropic.claude-sonnet-4-20250514-v1:0",
 	"claude-opus-4-20250514":     "anthropic.claude-opus-4-20250514-v1:0",
+	"claude-opus-4-1-20250805":   "anthropic.claude-opus-4-1-20250805-v1:0",
 }
 
 var awsModelCanCrossRegionMap = map[string]map[string]bool{
@@ -54,6 +55,9 @@ var awsModelCanCrossRegionMap = map[string]map[string]bool{
 	"anthropic.claude-opus-4-20250514-v1:0": {
 		"us": true,
 	},
+	"anthropic.claude-opus-4-1-20250805-v1:0": {
+		"us": true,
+	},
 }
 
 var awsRegionCrossModelPrefixMap = map[string]string{

relay/channel/aws/relay-aws.go

@@ -19,20 +19,31 @@ import (
 	"github.com/aws/aws-sdk-go-v2/credentials"
 	"github.com/aws/aws-sdk-go-v2/service/bedrockruntime"
 	bedrockruntimeTypes "github.com/aws/aws-sdk-go-v2/service/bedrockruntime/types"
+	"github.com/aws/smithy-go/auth/bearer"
 )
 
 func newAwsClient(c *gin.Context, info *relaycommon.RelayInfo) (*bedrockruntime.Client, error) {
 	awsSecret := strings.Split(info.ApiKey, "|")
-	if len(awsSecret) != 3 {
+	var client *bedrockruntime.Client
+	switch len(awsSecret) {
+	case 2:
+		apiKey := awsSecret[0]
+		region := awsSecret[1]
+		client = bedrockruntime.New(bedrockruntime.Options{
+			Region:                  region,
+			BearerAuthTokenProvider: bearer.StaticTokenProvider{Token: bearer.Token{Value: apiKey}},
+		})
+	case 3:
+		ak := awsSecret[0]
+		sk := awsSecret[1]
+		region := awsSecret[2]
+		client = bedrockruntime.New(bedrockruntime.Options{
+			Region:      region,
+			Credentials: aws.NewCredentialsCache(credentials.NewStaticCredentialsProvider(ak, sk, "")),
+		})
+	default:
 		return nil, errors.New("invalid aws secret key")
 	}
-	ak := awsSecret[0]
-	sk := awsSecret[1]
-	region := awsSecret[2]
-	client := bedrockruntime.New(bedrockruntime.Options{
-		Region:      region,
-		Credentials: aws.NewCredentialsCache(credentials.NewStaticCredentialsProvider(ak, sk, "")),
-	})
 
 	return client, nil
 }

relay/channel/baidu/relay-baidu.go

@@ -34,9 +34,9 @@ func requestOpenAI2Baidu(request dto.GeneralOpenAIRequest) *BaiduChatRequest {
 		EnableCitation: false,
 		UserId:         request.User,
 	}
-	if request.MaxTokens != 0 {
-		maxTokens := int(request.MaxTokens)
-		if request.MaxTokens == 1 {
+	if request.GetMaxTokens() != 0 {
+		maxTokens := int(request.GetMaxTokens())
+		if request.GetMaxTokens() == 1 {
 			maxTokens = 2
 		}
 		baiduRequest.MaxOutputTokens = &maxTokens

relay/channel/claude/adaptor.go

@@ -104,7 +104,7 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycom
 	if info.IsStream {
 		err, usage = ClaudeStreamHandler(c, resp, info, a.RequestMode)
 	} else {
-		err, usage = ClaudeHandler(c, resp, a.RequestMode, info)
+		err, usage = ClaudeHandler(c, resp, info, a.RequestMode)
 	}
 	return
 }

relay/channel/claude/constants.go

@@ -17,6 +17,8 @@ var ModelList = []string{
 	"claude-sonnet-4-20250514-thinking",
 	"claude-opus-4-20250514",
 	"claude-opus-4-20250514-thinking",
+	"claude-opus-4-1-20250805",
+	"claude-opus-4-1-20250805-thinking",
 }
 
 var ChannelName = "claude"

relay/channel/claude/relay-claude.go

@@ -149,7 +149,7 @@ func RequestOpenAI2ClaudeMessage(textRequest dto.GeneralOpenAIRequest) (*dto.Cla
 
 	claudeRequest := dto.ClaudeRequest{
 		Model:         textRequest.Model,
-		MaxTokens:     textRequest.MaxTokens,
+		MaxTokens:     textRequest.GetMaxTokens(),
 		StopSequences: nil,
 		Temperature:   textRequest.Temperature,
 		TopP:          textRequest.TopP,
@@ -740,7 +740,7 @@ func HandleClaudeResponseData(c *gin.Context, info *relaycommon.RelayInfo, claud
 	return nil
 }
 
-func ClaudeHandler(c *gin.Context, resp *http.Response, requestMode int, info *relaycommon.RelayInfo) (*types.NewAPIError, *dto.Usage) {
+func ClaudeHandler(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo, requestMode int) (*types.NewAPIError, *dto.Usage) {
 	defer common.CloseResponseBodyGracefully(resp)
 
 	claudeInfo := &ClaudeResponseInfo{

relay/channel/cloudflare/dto.go

@@ -5,7 +5,7 @@ import "one-api/dto"
 type CfRequest struct {
 	Messages    []dto.Message `json:"messages,omitempty"`
 	Lora        string        `json:"lora,omitempty"`
-	MaxTokens   int           `json:"max_tokens,omitempty"`
+	MaxTokens   uint          `json:"max_tokens,omitempty"`
 	Prompt      string        `json:"prompt,omitempty"`
 	Raw         bool          `json:"raw,omitempty"`
 	Stream      bool          `json:"stream,omitempty"`

relay/channel/cohere/dto.go

@@ -7,7 +7,7 @@ type CohereRequest struct {
 	ChatHistory []ChatHistory `json:"chat_history"`
 	Message     string        `json:"message"`
 	Stream      bool          `json:"stream"`
-	MaxTokens   int           `json:"max_tokens"`
+	MaxTokens   uint          `json:"max_tokens"`
 	SafetyMode  string        `json:"safety_mode,omitempty"`
 }
 

relay/channel/deepseek/adaptor.go

@@ -24,10 +24,9 @@ func (a *Adaptor) ConvertGeminiRequest(*gin.Context, *relaycommon.RelayInfo, *dt
 	return nil, errors.New("not implemented")
 }
 
-func (a *Adaptor) ConvertClaudeRequest(*gin.Context, *relaycommon.RelayInfo, *dto.ClaudeRequest) (any, error) {
-	//TODO implement me
-	panic("implement me")
-	return nil, nil
+func (a *Adaptor) ConvertClaudeRequest(c *gin.Context, info *relaycommon.RelayInfo, req *dto.ClaudeRequest) (any, error) {
+	adaptor := openai.Adaptor{}
+	return adaptor.ConvertClaudeRequest(c, info, req)
 }
 
 func (a *Adaptor) ConvertAudioRequest(c *gin.Context, info *relaycommon.RelayInfo, request dto.AudioRequest) (io.Reader, error) {

relay/channel/gemini/relay-gemini.go

@@ -49,12 +49,20 @@ const (
 	flash25LiteMaxBudget = 24576
 )
 
-// clampThinkingBudget 根据模型名称将预算限制在允许的范围内
-func clampThinkingBudget(modelName string, budget int) int {
-	isNew25Pro := strings.HasPrefix(modelName, "gemini-2.5-pro") &&
+func isNew25ProModel(modelName string) bool {
+	return strings.HasPrefix(modelName, "gemini-2.5-pro") &&
 		!strings.HasPrefix(modelName, "gemini-2.5-pro-preview-05-06") &&
 		!strings.HasPrefix(modelName, "gemini-2.5-pro-preview-03-25")
-	is25FlashLite := strings.HasPrefix(modelName, "gemini-2.5-flash-lite")
+}
+
+func is25FlashLiteModel(modelName string) bool {
+	return strings.HasPrefix(modelName, "gemini-2.5-flash-lite")
+}
+
+// clampThinkingBudget 根据模型名称将预算限制在允许的范围内
+func clampThinkingBudget(modelName string, budget int) int {
+	isNew25Pro := isNew25ProModel(modelName)
+	is25FlashLite := is25FlashLiteModel(modelName)
 
 	if is25FlashLite {
 		if budget < flash25LiteMinBudget {
@@ -81,7 +89,34 @@ func clampThinkingBudget(modelName string, budget int) int {
 	return budget
 }
 
-func ThinkingAdaptor(geminiRequest *dto.GeminiChatRequest, info *relaycommon.RelayInfo) {
+// "effort": "high" - Allocates a large portion of tokens for reasoning (approximately 80% of max_tokens)
+// "effort": "medium" - Allocates a moderate portion of tokens (approximately 50% of max_tokens)
+// "effort": "low" - Allocates a smaller portion of tokens (approximately 20% of max_tokens)
+func clampThinkingBudgetByEffort(modelName string, effort string) int {
+	isNew25Pro := isNew25ProModel(modelName)
+	is25FlashLite := is25FlashLiteModel(modelName)
+
+	maxBudget := 0
+	if is25FlashLite {
+		maxBudget = flash25LiteMaxBudget
+	}
+	if isNew25Pro {
+		maxBudget = pro25MaxBudget
+	} else {
+		maxBudget = flash25MaxBudget
+	}
+	switch effort {
+	case "high":
+		maxBudget = maxBudget * 80 / 100
+	case "medium":
+		maxBudget = maxBudget * 50 / 100
+	case "low":
+		maxBudget = maxBudget * 20 / 100
+	}
+	return clampThinkingBudget(modelName, maxBudget)
+}
+
+func ThinkingAdaptor(geminiRequest *dto.GeminiChatRequest, info *relaycommon.RelayInfo, oaiRequest ...dto.GeneralOpenAIRequest) {
 	if model_setting.GetGeminiSettings().ThinkingAdapterEnabled {
 		modelName := info.UpstreamModelName
 		isNew25Pro := strings.HasPrefix(modelName, "gemini-2.5-pro") &&
@@ -124,6 +159,11 @@ func ThinkingAdaptor(geminiRequest *dto.GeminiChatRequest, info *relaycommon.Rel
 					budgetTokens := model_setting.GetGeminiSettings().ThinkingAdapterBudgetTokensPercentage * float64(geminiRequest.GenerationConfig.MaxOutputTokens)
 					clampedBudget := clampThinkingBudget(modelName, int(budgetTokens))
 					geminiRequest.GenerationConfig.ThinkingConfig.ThinkingBudget = common.GetPointer(clampedBudget)
+				} else {
+					if len(oaiRequest) > 0 {
+						// 如果有reasoningEffort参数，则根据其值设置思考预算
+						geminiRequest.GenerationConfig.ThinkingConfig.ThinkingBudget = common.GetPointer(clampThinkingBudgetByEffort(modelName, oaiRequest[0].ReasoningEffort))
+					}
 				}
 			}
 		} else if strings.HasSuffix(modelName, "-nothinking") {
@@ -144,7 +184,7 @@ func CovertGemini2OpenAI(textRequest dto.GeneralOpenAIRequest, info *relaycommon
 		GenerationConfig: dto.GeminiChatGenerationConfig{
 			Temperature:     textRequest.Temperature,
 			TopP:            textRequest.TopP,
-			MaxOutputTokens: textRequest.MaxTokens,
+			MaxOutputTokens: textRequest.GetMaxTokens(),
 			Seed:            int64(textRequest.Seed),
 		},
 	}
@@ -156,7 +196,37 @@ func CovertGemini2OpenAI(textRequest dto.GeneralOpenAIRequest, info *relaycommon
 		}
 	}
 
-	ThinkingAdaptor(&geminiRequest, info)
+	adaptorWithExtraBody := false
+
+	if len(textRequest.ExtraBody) > 0 {
+		if !strings.HasSuffix(info.UpstreamModelName, "-nothinking") {
+			var extraBody map[string]interface{}
+			if err := common.Unmarshal(textRequest.ExtraBody, &extraBody); err != nil {
+				return nil, fmt.Errorf("invalid extra body: %w", err)
+			}
+			// eg. {"google":{"thinking_config":{"thinking_budget":5324,"include_thoughts":true}}}
+			if googleBody, ok := extraBody["google"].(map[string]interface{}); ok {
+				adaptorWithExtraBody = true
+				if thinkingConfig, ok := googleBody["thinking_config"].(map[string]interface{}); ok {
+					if budget, ok := thinkingConfig["thinking_budget"].(float64); ok {
+						budgetInt := int(budget)
+						geminiRequest.GenerationConfig.ThinkingConfig = &dto.GeminiThinkingConfig{
+							ThinkingBudget:  common.GetPointer(budgetInt),
+							IncludeThoughts: true,
+						}
+					} else {
+						geminiRequest.GenerationConfig.ThinkingConfig = &dto.GeminiThinkingConfig{
+							IncludeThoughts: true,
+						}
+					}
+				}
+			}
+		}
+	}
+
+	if !adaptorWithExtraBody {
+		ThinkingAdaptor(&geminiRequest, info, textRequest)
+	}
 
 	safetySettings := make([]dto.GeminiChatSafetySettings, 0, len(SafetySettingList))
 	for _, category := range SafetySettingList {

relay/channel/mistral/text.go

@@ -71,7 +71,7 @@ func requestOpenAI2Mistral(request *dto.GeneralOpenAIRequest) *dto.GeneralOpenAI
 		Messages:    messages,
 		Temperature: request.Temperature,
 		TopP:        request.TopP,
-		MaxTokens:   request.MaxTokens,
+		MaxTokens:   request.GetMaxTokens(),
 		Tools:       request.Tools,
 		ToolChoice:  request.ToolChoice,
 	}

relay/channel/ollama/relay-ollama.go

@@ -60,7 +60,7 @@ func requestOpenAI2Ollama(request *dto.GeneralOpenAIRequest) (*OllamaRequest, er
 		TopK:             request.TopK,
 		Stop:             Stop,
 		Tools:            request.Tools,
-		MaxTokens:        request.MaxTokens,
+		MaxTokens:        request.GetMaxTokens(),
 		ResponseFormat:   request.ResponseFormat,
 		FrequencyPenalty: request.FrequencyPenalty,
 		PresencePenalty:  request.PresencePenalty,

relay/channel/openai/adaptor.go

@@ -9,6 +9,7 @@ import (
 	"mime/multipart"
 	"net/http"
 	"net/textproto"
+	"one-api/common"
 	"one-api/constant"
 	"one-api/dto"
 	"one-api/relay/channel"
@@ -73,9 +74,6 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
-	if info.RelayFormat == relaycommon.RelayFormatClaude || info.RelayFormat == relaycommon.RelayFormatGemini {
-		return fmt.Sprintf("%s/v1/chat/completions", info.BaseUrl), nil
-	}
 	if info.RelayMode == relayconstant.RelayModeRealtime {
 		if strings.HasPrefix(info.BaseUrl, "https://") {
 			baseUrl := strings.TrimPrefix(info.BaseUrl, "https://")
@@ -122,6 +120,9 @@ func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 		url = strings.Replace(url, "{model}", info.UpstreamModelName, -1)
 		return url, nil
 	default:
+		if info.RelayFormat == relaycommon.RelayFormatClaude || info.RelayFormat == relaycommon.RelayFormatGemini {
+			return fmt.Sprintf("%s/v1/chat/completions", info.BaseUrl), nil
+		}
 		return relaycommon.GetFullRequestURL(info.BaseUrl, info.RequestURLPath, info.ChannelType), nil
 	}
 }
@@ -172,6 +173,23 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
 		if len(request.Usage) == 0 {
 			request.Usage = json.RawMessage(`{"include":true}`)
 		}
+		if strings.HasSuffix(info.UpstreamModelName, "-thinking") {
+			info.UpstreamModelName = strings.TrimSuffix(info.UpstreamModelName, "-thinking")
+			request.Model = info.UpstreamModelName
+			if len(request.Reasoning) == 0 {
+				reasoning := map[string]any{
+					"enabled": true,
+				}
+				if request.ReasoningEffort != "" {
+					reasoning["effort"] = request.ReasoningEffort
+				}
+				marshal, err := common.Marshal(reasoning)
+				if err != nil {
+					return nil, fmt.Errorf("error marshalling reasoning: %w", err)
+				}
+				request.Reasoning = marshal
+			}
+		}
 	}
 	if strings.HasPrefix(request.Model, "o") {
 		if request.MaxCompletionTokens == 0 && request.MaxTokens != 0 {

relay/channel/openai/relay_responses.go

@@ -37,9 +37,14 @@ func OaiResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http
 
 	// compute usage
 	usage := dto.Usage{}
-	usage.PromptTokens = responsesResponse.Usage.InputTokens
-	usage.CompletionTokens = responsesResponse.Usage.OutputTokens
-	usage.TotalTokens = responsesResponse.Usage.TotalTokens
+	if responsesResponse.Usage != nil {
+		usage.PromptTokens = responsesResponse.Usage.InputTokens
+		usage.CompletionTokens = responsesResponse.Usage.OutputTokens
+		usage.TotalTokens = responsesResponse.Usage.TotalTokens
+		if responsesResponse.Usage.InputTokensDetails != nil {
+			usage.PromptTokensDetails.CachedTokens = responsesResponse.Usage.InputTokensDetails.CachedTokens
+		}
+	}
 	// 解析 Tools 用量
 	for _, tool := range responsesResponse.Tools {
 		info.ResponsesUsageInfo.BuiltInTools[common.Interface2String(tool["type"])].CallCount++
@@ -64,9 +69,14 @@ func OaiResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
 			sendResponsesStreamData(c, streamResponse, data)
 			switch streamResponse.Type {
 			case "response.completed":
-				usage.PromptTokens = streamResponse.Response.Usage.InputTokens
-				usage.CompletionTokens = streamResponse.Response.Usage.OutputTokens
-				usage.TotalTokens = streamResponse.Response.Usage.TotalTokens
+				if streamResponse.Response.Usage != nil {
+					usage.PromptTokens = streamResponse.Response.Usage.InputTokens
+					usage.CompletionTokens = streamResponse.Response.Usage.OutputTokens
+					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)

relay/channel/palm/relay-palm.go

@@ -18,30 +18,6 @@ import (
 // https://developers.generativeai.google/api/rest/generativelanguage/models/generateMessage#request-body
 // https://developers.generativeai.google/api/rest/generativelanguage/models/generateMessage#response-body
 
-func requestOpenAI2PaLM(textRequest dto.GeneralOpenAIRequest) *PaLMChatRequest {
-	palmRequest := PaLMChatRequest{
-		Prompt: PaLMPrompt{
-			Messages: make([]PaLMChatMessage, 0, len(textRequest.Messages)),
-		},
-		Temperature:    textRequest.Temperature,
-		CandidateCount: textRequest.N,
-		TopP:           textRequest.TopP,
-		TopK:           textRequest.MaxTokens,
-	}
-	for _, message := range textRequest.Messages {
-		palmMessage := PaLMChatMessage{
-			Content: message.StringContent(),
-		}
-		if message.Role == "user" {
-			palmMessage.Author = "0"
-		} else {
-			palmMessage.Author = "1"
-		}
-		palmRequest.Prompt.Messages = append(palmRequest.Prompt.Messages, palmMessage)
-	}
-	return &palmRequest
-}
-
 func responsePaLM2OpenAI(response *PaLMChatResponse) *dto.OpenAITextResponse {
 	fullTextResponse := dto.OpenAITextResponse{
 		Choices: make([]dto.OpenAITextResponseChoice, 0, len(response.Candidates)),

relay/channel/perplexity/relay-perplexity.go

@@ -16,6 +16,6 @@ func requestOpenAI2Perplexity(request dto.GeneralOpenAIRequest) *dto.GeneralOpen
 		Messages:    messages,
 		Temperature: request.Temperature,
 		TopP:        request.TopP,
-		MaxTokens:   request.MaxTokens,
+		MaxTokens:   request.GetMaxTokens(),
 	}
 }

relay/channel/vertex/adaptor.go

@@ -35,6 +35,7 @@ var claudeModelMap = map[string]string{
 	"claude-3-7-sonnet-20250219": "claude-3-7-sonnet@20250219",
 	"claude-sonnet-4-20250514":   "claude-sonnet-4@20250514",
 	"claude-opus-4-20250514":     "claude-opus-4@20250514",
+	"claude-opus-4-1-20250805":   "claude-opus-4-1@20250805",
 }
 
 const anthropicVersion = "vertex-2023-10-16"
@@ -237,7 +238,7 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycom
 	} else {
 		switch a.RequestMode {
 		case RequestModeClaude:
-			err, usage = claude.ClaudeHandler(c, resp, claude.RequestModeMessage, info)
+			err, usage = claude.ClaudeHandler(c, resp, info, claude.RequestModeMessage)
 		case RequestModeGemini:
 			if info.RelayMode == constant.RelayModeGemini {
 				usage, err = gemini.GeminiTextGenerationHandler(c, info, resp)

relay/channel/xunfei/relay-xunfei.go

@@ -48,7 +48,7 @@ func requestOpenAI2Xunfei(request dto.GeneralOpenAIRequest, xunfeiAppId string,
 	xunfeiRequest.Parameter.Chat.Domain = domain
 	xunfeiRequest.Parameter.Chat.Temperature = request.Temperature
 	xunfeiRequest.Parameter.Chat.TopK = request.N
-	xunfeiRequest.Parameter.Chat.MaxTokens = request.MaxTokens
+	xunfeiRequest.Parameter.Chat.MaxTokens = request.GetMaxTokens()
 	xunfeiRequest.Payload.Message.Text = messages
 	return &xunfeiRequest
 }

relay/channel/zhipu_4v/relay-zhipu_v4.go

@@ -105,7 +105,7 @@ func requestOpenAI2Zhipu(request dto.GeneralOpenAIRequest) *dto.GeneralOpenAIReq
 		Messages:    messages,
 		Temperature: request.Temperature,
 		TopP:        request.TopP,
-		MaxTokens:   request.MaxTokens,
+		MaxTokens:   request.GetMaxTokens(),
 		Stop:        Stop,
 		Tools:       request.Tools,
 		ToolChoice:  request.ToolChoice,

relay/common/relay_info.go

@@ -225,6 +225,9 @@ func GenRelayInfo(c *gin.Context) *RelayInfo {
 	userId := common.GetContextKeyInt(c, constant.ContextKeyUserId)
 	tokenUnlimited := common.GetContextKeyBool(c, constant.ContextKeyTokenUnlimited)
 	startTime := common.GetContextKeyTime(c, constant.ContextKeyRequestStartTime)
+	if startTime.IsZero() {
+		startTime = time.Now()
+	}
 	// firstResponseTime = time.Now() - 1 second
 
 	apiType, _ := common.ChannelType2APIType(channelType)

service/convert.go

@@ -283,7 +283,9 @@ func StreamResponseOpenAI2Claude(openAIResponse *dto.ChatCompletionsStreamRespon
 		if chosenChoice.FinishReason != nil && *chosenChoice.FinishReason != "" {
 			// should be done
 			info.FinishReason = *chosenChoice.FinishReason
-			return claudeResponses
+			if !info.Done {
+				return claudeResponses
+			}
 		}
 		if info.Done {
 			claudeResponses = append(claudeResponses, generateStopBlock(info.ClaudeConvertInfo.Index))
@@ -432,6 +434,8 @@ func stopReasonOpenAI2Claude(reason string) string {
 		return "end_turn"
 	case "stop_sequence":
 		return "stop_sequence"
+	case "length":
+		fallthrough
 	case "max_tokens":
 		return "max_tokens"
 	case "tool_calls":

service/error.go

@@ -93,6 +93,9 @@ func RelayErrorHandler(resp *http.Response, showBodyWhenFail bool) (newApiErr *t
 		if showBodyWhenFail {
 			newApiErr.Err = fmt.Errorf("bad response status code %d, body: %s", resp.StatusCode, string(responseBody))
 		} else {
+			if common.DebugEnabled {
+				println(fmt.Sprintf("bad response status code %d, body: %s", resp.StatusCode, string(responseBody)))
+			}
 			newApiErr.Err = fmt.Errorf("bad response status code %d", resp.StatusCode)
 		}
 		return

setting/ratio_setting/cache_ratio.go

@@ -40,6 +40,8 @@ var defaultCacheRatio = map[string]float64{
 	"claude-sonnet-4-20250514-thinking":   0.1,
 	"claude-opus-4-20250514":              0.1,
 	"claude-opus-4-20250514-thinking":     0.1,
+	"claude-opus-4-1-20250805":            0.1,
+	"claude-opus-4-1-20250805-thinking":   0.1,
 }
 
 var defaultCreateCacheRatio = map[string]float64{
@@ -55,6 +57,8 @@ var defaultCreateCacheRatio = map[string]float64{
 	"claude-sonnet-4-20250514-thinking":   1.25,
 	"claude-opus-4-20250514":              1.25,
 	"claude-opus-4-20250514-thinking":     1.25,
+	"claude-opus-4-1-20250805":            1.25,
+	"claude-opus-4-1-20250805-thinking":   1.25,
 }
 
 //var defaultCreateCacheRatio = map[string]float64{}

setting/ratio_setting/model_ratio.go

@@ -118,6 +118,7 @@ var defaultModelRatio = map[string]float64{
 	"claude-sonnet-4-20250514":                  1.5,
 	"claude-3-opus-20240229":                    7.5, // $15 / 1M tokens
 	"claude-opus-4-20250514":                    7.5,
+	"claude-opus-4-1-20250805":                  7.5,
 	"ERNIE-4.0-8K":                              0.120 * RMB,
 	"ERNIE-3.5-8K":                              0.012 * RMB,
 	"ERNIE-3.5-8K-0205":                         0.024 * RMB,
@@ -334,12 +335,8 @@ func GetModelPrice(name string, printErr bool) (float64, bool) {
 	modelPriceMapMutex.RLock()
 	defer modelPriceMapMutex.RUnlock()
 
-	if strings.HasPrefix(name, "gpt-4-gizmo") {
-		name = "gpt-4-gizmo-*"
-	}
-	if strings.HasPrefix(name, "gpt-4o-gizmo") {
-		name = "gpt-4o-gizmo-*"
-	}
+	name = FormatMatchingModelName(name)
+
 	price, ok := modelPriceMap[name]
 	if !ok {
 		if printErr {
@@ -373,11 +370,8 @@ func GetModelRatio(name string) (float64, bool, string) {
 	modelRatioMapMutex.RLock()
 	defer modelRatioMapMutex.RUnlock()
 
-	name = handleThinkingBudgetModel(name, "gemini-2.5-flash", "gemini-2.5-flash-thinking-*")
-	name = handleThinkingBudgetModel(name, "gemini-2.5-pro", "gemini-2.5-pro-thinking-*")
-	if strings.HasPrefix(name, "gpt-4-gizmo") {
-		name = "gpt-4-gizmo-*"
-	}
+	name = FormatMatchingModelName(name)
+
 	ratio, ok := modelRatioMap[name]
 	if !ok {
 		return 37.5, operation_setting.SelfUseModeEnabled, name
@@ -428,12 +422,9 @@ func UpdateCompletionRatioByJSONString(jsonStr string) error {
 func GetCompletionRatio(name string) float64 {
 	CompletionRatioMutex.RLock()
 	defer CompletionRatioMutex.RUnlock()
-	if strings.HasPrefix(name, "gpt-4-gizmo") {
-		name = "gpt-4-gizmo-*"
-	}
-	if strings.HasPrefix(name, "gpt-4o-gizmo") {
-		name = "gpt-4o-gizmo-*"
-	}
+
+	name = FormatMatchingModelName(name)
+
 	if strings.Contains(name, "/") {
 		if ratio, ok := CompletionRatio[name]; ok {
 			return ratio
@@ -663,3 +654,16 @@ func GetCompletionRatioCopy() map[string]float64 {
 	}
 	return copyMap
 }
+
+// 转换模型名，减少渠道必须配置各种带参数模型
+func FormatMatchingModelName(name string) string {
+	name = handleThinkingBudgetModel(name, "gemini-2.5-flash", "gemini-2.5-flash-thinking-*")
+	name = handleThinkingBudgetModel(name, "gemini-2.5-pro", "gemini-2.5-pro-thinking-*")
+	if strings.HasPrefix(name, "gpt-4-gizmo") {
+		name = "gpt-4-gizmo-*"
+	}
+	if strings.HasPrefix(name, "gpt-4o-gizmo") {
+		name = "gpt-4o-gizmo-*"
+	}
+	return name
+}

types/error.go

@@ -189,9 +189,13 @@ func NewError(err error, errorCode ErrorCode, ops ...NewAPIErrorOptions) *NewAPI
 }
 
 func NewOpenAIError(err error, errorCode ErrorCode, statusCode int, ops ...NewAPIErrorOptions) *NewAPIError {
+	if errorCode == ErrorCodeDoRequestFailed {
+		err = errors.New("upstream error: do request failed")
+	}
 	openaiError := OpenAIError{
 		Message: err.Error(),
 		Type:    string(errorCode),
+		Code:    errorCode,
 	}
 	return WithOpenAIError(openaiError, statusCode, ops...)
 }
@@ -199,6 +203,7 @@ func NewOpenAIError(err error, errorCode ErrorCode, statusCode int, ops ...NewAP
 func InitOpenAIError(errorCode ErrorCode, statusCode int, ops ...NewAPIErrorOptions) *NewAPIError {
 	openaiError := OpenAIError{
 		Type: string(errorCode),
+		Code: errorCode,
 	}
 	return WithOpenAIError(openaiError, statusCode, ops...)
 }
@@ -224,7 +229,11 @@ func NewErrorWithStatusCode(err error, errorCode ErrorCode, statusCode int, ops
 func WithOpenAIError(openAIError OpenAIError, statusCode int, ops ...NewAPIErrorOptions) *NewAPIError {
 	code, ok := openAIError.Code.(string)
 	if !ok {
-		code = fmt.Sprintf("%v", openAIError.Code)
+		if openAIError.Code == nil {
+			code = fmt.Sprintf("%v", openAIError.Code)
+		} else {
+			code = "unknown_error"
+		}
 	}
 	if openAIError.Type == "" {
 		openAIError.Type = "upstream_error"

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

@@ -395,8 +395,7 @@ const MultiKeyManageModal = ({
       }
       visible={visible}
       onCancel={onCancel}
-      width={800}
-      height={600}
+      width={900}
       footer={
         <Space>
           <Button onClick={onCancel}>{t('关闭')}</Button>
@@ -452,11 +451,11 @@ const MultiKeyManageModal = ({
         </Space>
       }
     >
-      <div style={{ padding: '16px 0' }}>
+      <div style={{ height: '100%', display: 'flex', flexDirection: 'column' }}>
         {/* Statistics Banner */}
         <Banner
           type='info'
-          style={{ marginBottom: '16px' }}
+          style={{ marginBottom: '16px', flexShrink: 0 }}
           description={
             <div>
               <Text>
@@ -479,7 +478,7 @@ const MultiKeyManageModal = ({
         />
 
         {/* Filter Controls */}
-        <div style={{ marginBottom: '16px', display: 'flex', alignItems: 'center', gap: '12px' }}>
+        <div style={{ marginBottom: '16px', display: 'flex', alignItems: 'center', gap: '12px', flexShrink: 0 }}>
           <Text style={{ fontSize: '14px', fontWeight: '500' }}>{t('状态筛选')}:</Text>
           <Select
             value={statusFilter}
@@ -501,75 +500,87 @@ const MultiKeyManageModal = ({
         </div>
 
         {/* Key Status Table */}
-        <Spin spinning={loading}>
-          {keyStatusList.length > 0 ? (
-            <>
-              <Table
-                columns={columns}
-                dataSource={keyStatusList}
-                pagination={false}
-                size='small'
-                bordered
-                rowKey='index'
-                style={{ marginBottom: '16px' }}
-              />
-              
-              {/* Pagination */}
-              {total > 0 && (
-                <div style={{ display: 'flex', justifyContent: 'space-between', alignItems: 'center' }}>
-                  <Text type='quaternary' style={{ fontSize: '12px' }}>
-                    {t('显示第 {{start}}-{{end}} 条，共 {{total}} 条', {
-                      start: (currentPage - 1) * pageSize + 1,
-                      end: Math.min(currentPage * pageSize, total),
-                      total: total
-                    })}
-                  </Text>
-                  
-                  <div style={{ display: 'flex', alignItems: 'center', gap: '12px' }}>
-                    <Text type='quaternary' style={{ fontSize: '12px' }}>
-                      {t('每页显示')}:
-                    </Text>
-                    <Select
-                      value={pageSize}
-                      onChange={handlePageSizeChange}
-                      size='small'
-                      style={{ width: '80px' }}
-                    >
-                      <Select.Option value={50}>50</Select.Option>
-                      <Select.Option value={100}>100</Select.Option>
-                      <Select.Option value={500}>500</Select.Option>
-                      <Select.Option value={1000}>1000</Select.Option>
-                    </Select>
-                    
-                    <Pagination
-                      current={currentPage}
-                      total={total}
-                      pageSize={pageSize}
-                      showSizeChanger={false}
-                      showQuickJumper
-                      size='small'
-                      onChange={handlePageChange}
-                      showTotal={(total, range) => 
-                        t('第 {{current}} / {{total}} 页', {
-                          current: currentPage,
-                          total: totalPages
-                        })
-                      }
-                    />
-                  </div>
+        <div style={{ flex: 1, display: 'flex', flexDirection: 'column', minHeight: 0 }}>
+          <Spin spinning={loading}>
+            {keyStatusList.length > 0 ? (
+              <div style={{ height: '100%', display: 'flex', flexDirection: 'column' }}>
+                <div style={{ flex: 1, overflow: 'auto', marginBottom: '16px' }}>
+                  <Table
+                    columns={columns}
+                    dataSource={keyStatusList}
+                    pagination={false}
+                    size='small'
+                    bordered
+                    rowKey='index'
+                    scroll={{ y: 'calc(100vh - 400px)' }}
+                  />
                 </div>
-              )}
-            </>
-          ) : (
-            !loading && (
-              <Empty
-                image={Empty.PRESENTED_IMAGE_SIMPLE}
-                title={t('暂无密钥数据')}
-                description={t('请检查渠道配置或刷新重试')}
-              />
-            )
-          )}
-        </Spin>
+                
+                {/* Pagination */}
+                {total > 0 && (
+                  <div style={{ 
+                    display: 'flex', 
+                    justifyContent: 'space-between', 
+                    alignItems: 'center',
+                    flexShrink: 0,
+                    padding: '12px 0',
+                    borderTop: '1px solid var(--semi-color-border)',
+                    backgroundColor: 'var(--semi-color-bg-1)'
+                  }}>
+                    <Text type='quaternary' style={{ fontSize: '12px' }}>
+                      {t('显示第 {{start}}-{{end}} 条，共 {{total}} 条', {
+                        start: (currentPage - 1) * pageSize + 1,
+                        end: Math.min(currentPage * pageSize, total),
+                        total: total
+                      })}
+                    </Text>
+                    
+                    <div style={{ display: 'flex', alignItems: 'center', gap: '12px' }}>
+                      <Text type='quaternary' style={{ fontSize: '12px' }}>
+                        {t('每页显示')}:
+                      </Text>
+                      <Select
+                        value={pageSize}
+                        onChange={handlePageSizeChange}
+                        size='small'
+                        style={{ width: '80px' }}
+                      >
+                        <Select.Option value={50}>50</Select.Option>
+                        <Select.Option value={100}>100</Select.Option>
+                        <Select.Option value={500}>500</Select.Option>
+                        <Select.Option value={1000}>1000</Select.Option>
+                      </Select>
+                      
+                      <Pagination
+                        current={currentPage}
+                        total={total}
+                        pageSize={pageSize}
+                        showSizeChanger={false}
+                        showQuickJumper
+                        size='small'
+                        onChange={handlePageChange}
+                        showTotal={(total, range) => 
+                          t('第 {{current}} / {{total}} 页', {
+                            current: currentPage,
+                            total: totalPages
+                          })
+                        }
+                      />
+                    </div>
+                  </div>
+                )}
+              </div>
+            ) : (
+              !loading && (
+                <Empty
+                  image={Empty.PRESENTED_IMAGE_SIMPLE}
+                  title={t('暂无密钥数据')}
+                  description={t('请检查渠道配置或刷新重试')}
+                />
+              )
+            )}
+          </Spin>
+        </div>
       </div>
     </Modal>
   );

web/src/helpers/render.js

@@ -1156,6 +1156,7 @@ export function renderLogContent(
   modelPrice = -1,
   groupRatio,
   user_group_ratio,
+  cacheRatio = 1.0,
   image = false,
   imageRatio = 1.0,
   webSearch = false,
@@ -1174,9 +1175,10 @@ export function renderLogContent(
   } else {
     if (image) {
       return i18next.t(
-        '模型倍率 {{modelRatio}}，输出倍率 {{completionRatio}}，图片输入倍率 {{imageRatio}}，{{ratioType}} {{ratio}}',
+        '模型倍率 {{modelRatio}}，缓存倍率 {{cacheRatio}}，输出倍率 {{completionRatio}}，图片输入倍率 {{imageRatio}}，{{ratioType}} {{ratio}}',
         {
           modelRatio: modelRatio,
+          cacheRatio: cacheRatio,
           completionRatio: completionRatio,
           imageRatio: imageRatio,
           ratioType: ratioLabel,
@@ -1185,9 +1187,10 @@ export function renderLogContent(
       );
     } else if (webSearch) {
       return i18next.t(
-        '模型倍率 {{modelRatio}}，输出倍率 {{completionRatio}}，{{ratioType}} {{ratio}}，Web 搜索调用 {{webSearchCallCount}} 次',
+        '模型倍率 {{modelRatio}}，缓存倍率 {{cacheRatio}}，输出倍率 {{completionRatio}}，{{ratioType}} {{ratio}}，Web 搜索调用 {{webSearchCallCount}} 次',
         {
           modelRatio: modelRatio,
+          cacheRatio: cacheRatio,
           completionRatio: completionRatio,
           ratioType: ratioLabel,
           ratio,
@@ -1196,9 +1199,10 @@ export function renderLogContent(
       );
     } else {
       return i18next.t(
-        '模型倍率 {{modelRatio}}，输出倍率 {{completionRatio}}，{{ratioType}} {{ratio}}',
+        '模型倍率 {{modelRatio}}，缓存倍率 {{cacheRatio}}，输出倍率 {{completionRatio}}，{{ratioType}} {{ratio}}',
         {
           modelRatio: modelRatio,
+          cacheRatio: cacheRatio,
           completionRatio: completionRatio,
           ratioType: ratioLabel,
           ratio,

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

@@ -366,6 +366,7 @@ export const useLogsData = () => {
               other.model_price,
               other.group_ratio,
               other?.user_group_ratio,
+              other.cache_ratio || 1.0,
               false,
               1.0,
               other.web_search || false,

web/src/pages/Setting/Ratio/ModelSettingsVisualEditor.js

@@ -44,6 +44,7 @@ export default function ModelSettingsVisualEditor(props) {
   const { t } = useTranslation();
   const [models, setModels] = useState([]);
   const [visible, setVisible] = useState(false);
+  const [isEditMode, setIsEditMode] = useState(false);
   const [currentModel, setCurrentModel] = useState(null);
   const [searchText, setSearchText] = useState('');
   const [currentPage, setCurrentPage] = useState(1);
@@ -386,9 +387,11 @@ export default function ModelSettingsVisualEditor(props) {
     setCurrentModel(null);
     setPricingMode('per-token');
     setPricingSubMode('ratio');
+    setIsEditMode(false);
   };
 
   const editModel = (record) => {
+    setIsEditMode(true);
     // Determine which pricing mode to use based on the model's current configuration
     let initialPricingMode = 'per-token';
     let initialPricingSubMode = 'ratio';
@@ -500,13 +503,7 @@ export default function ModelSettingsVisualEditor(props) {
       </Space>
 
       <Modal
-        title={
-          currentModel &&
-            currentModel.name &&
-            models.some((model) => model.name === currentModel.name)
-            ? t('编辑模型')
-            : t('添加模型')
-        }
+        title={isEditMode ? t('编辑模型') : t('添加模型')}
         visible={visible}
         onCancel={() => {
           resetModalState();
@@ -562,11 +559,7 @@ export default function ModelSettingsVisualEditor(props) {
             label={t('模型名称')}
             placeholder='strawberry'
             required
-            disabled={
-              currentModel &&
-              currentModel.name &&
-              models.some((model) => model.name === currentModel.name)
-            }
+            disabled={isEditMode}
             onChange={(value) =>
               setCurrentModel((prev) => ({ ...prev, name: value }))
             }