feat: change ParallelToolCalls and Store fields to json.RawMessage type

feat: 支持 gemini-embedding-001

constant/task.go

@@ -11,8 +11,10 @@ const (
 	SunoActionMusic  = "MUSIC"
 	SunoActionLyrics = "LYRICS"
 
-	TaskActionGenerate     = "generate"
-	TaskActionTextGenerate = "textGenerate"
+	TaskActionGenerate          = "generate"
+	TaskActionTextGenerate      = "textGenerate"
+	TaskActionFirstTailGenerate = "firstTailGenerate"
+	TaskActionReferenceGenerate = "referenceGenerate"
 )
 
 var SunoModel2Action = map[string]string{

controller/channel.go

@@ -501,9 +501,10 @@ func validateChannel(channel *model.Channel, isAdd bool) error {
 }
 
 type AddChannelRequest struct {
-	Mode         string                `json:"mode"`
-	MultiKeyMode constant.MultiKeyMode `json:"multi_key_mode"`
-	Channel      *model.Channel        `json:"channel"`
+	Mode                      string                `json:"mode"`
+	MultiKeyMode              constant.MultiKeyMode `json:"multi_key_mode"`
+	BatchAddSetKeyPrefix2Name bool                  `json:"batch_add_set_key_prefix_2_name"`
+	Channel                   *model.Channel        `json:"channel"`
 }
 
 func getVertexArrayKeys(keys string) ([]string, error) {
@@ -616,6 +617,13 @@ func AddChannel(c *gin.Context) {
 		}
 		localChannel := addChannelRequest.Channel
 		localChannel.Key = key
+		if addChannelRequest.BatchAddSetKeyPrefix2Name && len(keys) > 1 {
+			keyPrefix := localChannel.Key
+			if len(localChannel.Key) > 8 {
+				keyPrefix = localChannel.Key[:8]
+			}
+			localChannel.Name = fmt.Sprintf("%s %s", localChannel.Name, keyPrefix)
+		}
 		channels = append(channels, *localChannel)
 	}
 	err = model.BatchInsertChannels(channels)

controller/option.go

@@ -128,6 +128,33 @@ func UpdateOption(c *gin.Context) {
 			})
 			return
 		}
+	case "ImageRatio":
+		err = ratio_setting.UpdateImageRatioByJSONString(option.Value.(string))
+		if err != nil {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "图片倍率设置失败: " + err.Error(),
+			})
+			return
+		}
+	case "AudioRatio":
+		err = ratio_setting.UpdateAudioRatioByJSONString(option.Value.(string))
+		if err != nil {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "音频倍率设置失败: " + err.Error(),
+			})
+			return
+		}
+	case "AudioCompletionRatio":
+		err = ratio_setting.UpdateAudioCompletionRatioByJSONString(option.Value.(string))
+		if err != nil {
+			c.JSON(http.StatusOK, gin.H{
+				"success": false,
+				"message": "音频补全倍率设置失败: " + err.Error(),
+			})
+			return
+		}
 	case "ModelRequestRateLimitGroup":
 		err = setting.CheckModelRequestRateLimitGroup(option.Value.(string))
 		if err != nil {

dto/openai_request.go

@@ -772,11 +772,12 @@ type OpenAIResponsesRequest struct {
 	Instructions       json.RawMessage `json:"instructions,omitempty"`
 	MaxOutputTokens    uint            `json:"max_output_tokens,omitempty"`
 	Metadata           json.RawMessage `json:"metadata,omitempty"`
-	ParallelToolCalls  bool            `json:"parallel_tool_calls,omitempty"`
+	ParallelToolCalls  json.RawMessage `json:"parallel_tool_calls,omitempty"`
 	PreviousResponseID string          `json:"previous_response_id,omitempty"`
 	Reasoning          *Reasoning      `json:"reasoning,omitempty"`
 	ServiceTier        string          `json:"service_tier,omitempty"`
-	Store              bool            `json:"store,omitempty"`
+	Store              json.RawMessage `json:"store,omitempty"`
+	PromptCacheKey     json.RawMessage `json:"prompt_cache_key,omitempty"`
 	Stream             bool            `json:"stream,omitempty"`
 	Temperature        float64         `json:"temperature,omitempty"`
 	Text               json.RawMessage `json:"text,omitempty"`

model/option.go

@@ -112,6 +112,9 @@ func InitOptionMap() {
 	common.OptionMap["GroupGroupRatio"] = ratio_setting.GroupGroupRatio2JSONString()
 	common.OptionMap["UserUsableGroups"] = setting.UserUsableGroups2JSONString()
 	common.OptionMap["CompletionRatio"] = ratio_setting.CompletionRatio2JSONString()
+	common.OptionMap["ImageRatio"] = ratio_setting.ImageRatio2JSONString()
+	common.OptionMap["AudioRatio"] = ratio_setting.AudioRatio2JSONString()
+	common.OptionMap["AudioCompletionRatio"] = ratio_setting.AudioCompletionRatio2JSONString()
 	common.OptionMap["TopUpLink"] = common.TopUpLink
 	//common.OptionMap["ChatLink"] = common.ChatLink
 	//common.OptionMap["ChatLink2"] = common.ChatLink2
@@ -397,6 +400,12 @@ func updateOptionMap(key string, value string) (err error) {
 		err = ratio_setting.UpdateModelPriceByJSONString(value)
 	case "CacheRatio":
 		err = ratio_setting.UpdateCacheRatioByJSONString(value)
+	case "ImageRatio":
+		err = ratio_setting.UpdateImageRatioByJSONString(value)
+	case "AudioRatio":
+		err = ratio_setting.UpdateAudioRatioByJSONString(value)
+	case "AudioCompletionRatio":
+		err = ratio_setting.UpdateAudioCompletionRatioByJSONString(value)
 	case "TopUpLink":
 		common.TopUpLink = value
 	//case "ChatLink":

relay/channel/deepseek/adaptor.go

@@ -3,17 +3,17 @@ package deepseek
 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"
 	"strings"
-
-	"github.com/gin-gonic/gin"
 )
 
 type Adaptor struct {
@@ -25,7 +25,7 @@ func (a *Adaptor) ConvertGeminiRequest(*gin.Context, *relaycommon.RelayInfo, *dt
 }
 
 func (a *Adaptor) ConvertClaudeRequest(c *gin.Context, info *relaycommon.RelayInfo, req *dto.ClaudeRequest) (any, error) {
-	adaptor := openai.Adaptor{}
+	adaptor := claude.Adaptor{}
 	return adaptor.ConvertClaudeRequest(c, info, req)
 }
 
@@ -44,14 +44,19 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
 
 func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	fimBaseUrl := info.ChannelBaseUrl
-	if !strings.HasSuffix(info.ChannelBaseUrl, "/beta") {
-		fimBaseUrl += "/beta"
-	}
-	switch info.RelayMode {
-	case constant.RelayModeCompletions:
-		return fmt.Sprintf("%s/completions", fimBaseUrl), nil
+	switch info.RelayFormat {
+	case types.RelayFormatClaude:
+		return fmt.Sprintf("%s/anthropic/v1/messages", info.ChannelBaseUrl), nil
 	default:
-		return fmt.Sprintf("%s/v1/chat/completions", info.ChannelBaseUrl), nil
+		if !strings.HasSuffix(info.ChannelBaseUrl, "/beta") {
+			fimBaseUrl += "/beta"
+		}
+		switch info.RelayMode {
+		case constant.RelayModeCompletions:
+			return fmt.Sprintf("%s/completions", fimBaseUrl), nil
+		default:
+			return fmt.Sprintf("%s/v1/chat/completions", info.ChannelBaseUrl), nil
+		}
 	}
 }
 
@@ -87,12 +92,17 @@ 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) {
-	if info.IsStream {
-		usage, err = openai.OaiStreamHandler(c, info, resp)
-	} else {
-		usage, err = openai.OpenaiHandler(c, info, resp)
+	switch info.RelayFormat {
+	case types.RelayFormatClaude:
+		if info.IsStream {
+			return claude.ClaudeStreamHandler(c, resp, info, claude.RequestModeMessage)
+		} else {
+			return claude.ClaudeHandler(c, resp, info, claude.RequestModeMessage)
+		}
+	default:
+		adaptor := openai.Adaptor{}
+		return adaptor.DoResponse(c, resp, info)
 	}
-	return
 }
 
 func (a *Adaptor) GetModelList() []string {

relay/channel/gemini/adaptor.go

@@ -215,8 +215,8 @@ 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) {
 	if info.RelayMode == constant.RelayModeGemini {
-		if strings.HasSuffix(info.RequestURLPath, ":embedContent") ||
-			strings.HasSuffix(info.RequestURLPath, ":batchEmbedContents") {
+		if strings.Contains(info.RequestURLPath, ":embedContent") ||
+			strings.Contains(info.RequestURLPath, ":batchEmbedContents") {
 			return NativeGeminiEmbeddingHandler(c, resp, info)
 		}
 		if info.IsStream {

relay/channel/gemini/relay-gemini.go

@@ -23,6 +23,7 @@ import (
 	"github.com/gin-gonic/gin"
 )
 
+// https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/inference?hl=zh-cn#blob
 var geminiSupportedMimeTypes = map[string]bool{
 	"application/pdf": true,
 	"audio/mpeg":      true,
@@ -30,6 +31,7 @@ var geminiSupportedMimeTypes = map[string]bool{
 	"audio/wav":       true,
 	"image/png":       true,
 	"image/jpeg":      true,
+	"image/webp":      true,
 	"text/plain":      true,
 	"video/mov":       true,
 	"video/mpeg":      true,

relay/channel/moonshot/adaptor.go

@@ -25,7 +25,7 @@ func (a *Adaptor) ConvertGeminiRequest(*gin.Context, *relaycommon.RelayInfo, *dt
 }
 
 func (a *Adaptor) ConvertClaudeRequest(c *gin.Context, info *relaycommon.RelayInfo, req *dto.ClaudeRequest) (any, error) {
-	adaptor := openai.Adaptor{}
+	adaptor := claude.Adaptor{}
 	return adaptor.ConvertClaudeRequest(c, info, req)
 }
 

relay/channel/task/jimeng/adaptor.go

@@ -94,6 +94,9 @@ func (a *TaskAdaptor) ValidateRequestAndSetAction(c *gin.Context, info *relaycom
 
 // BuildRequestURL constructs the upstream URL.
 func (a *TaskAdaptor) BuildRequestURL(info *relaycommon.RelayInfo) (string, error) {
+	if isNewAPIRelay(info.ApiKey) {
+		return fmt.Sprintf("%s/jimeng/?Action=CVSync2AsyncSubmitTask&Version=2022-08-31", a.baseURL), nil
+	}
 	return fmt.Sprintf("%s/?Action=CVSync2AsyncSubmitTask&Version=2022-08-31", a.baseURL), nil
 }
 
@@ -101,7 +104,12 @@ func (a *TaskAdaptor) BuildRequestURL(info *relaycommon.RelayInfo) (string, erro
 func (a *TaskAdaptor) BuildRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
 	req.Header.Set("Content-Type", "application/json")
 	req.Header.Set("Accept", "application/json")
-	return a.signRequest(req, a.accessKey, a.secretKey)
+	if isNewAPIRelay(info.ApiKey) {
+		req.Header.Set("Authorization", "Bearer "+info.ApiKey)
+	} else {
+		return a.signRequest(req, a.accessKey, a.secretKey)
+	}
+	return nil
 }
 
 // BuildRequestBody converts request into Jimeng specific format.
@@ -161,6 +169,9 @@ func (a *TaskAdaptor) FetchTask(baseUrl, key string, body map[string]any) (*http
 	}
 
 	uri := fmt.Sprintf("%s/?Action=CVSync2AsyncGetResult&Version=2022-08-31", baseUrl)
+	if isNewAPIRelay(key) {
+		uri = fmt.Sprintf("%s/jimeng/?Action=CVSync2AsyncGetResult&Version=2022-08-31", a.baseURL)
+	}
 	payload := map[string]string{
 		"req_key": "jimeng_vgfm_t2v_l20", // This is fixed value from doc: https://www.volcengine.com/docs/85621/1544774
 		"task_id": taskID,
@@ -178,17 +189,20 @@ func (a *TaskAdaptor) FetchTask(baseUrl, key string, body map[string]any) (*http
 	req.Header.Set("Accept", "application/json")
 	req.Header.Set("Content-Type", "application/json")
 
-	keyParts := strings.Split(key, "|")
-	if len(keyParts) != 2 {
-		return nil, fmt.Errorf("invalid api key format for jimeng: expected 'ak|sk'")
-	}
-	accessKey := strings.TrimSpace(keyParts[0])
-	secretKey := strings.TrimSpace(keyParts[1])
+	if isNewAPIRelay(key) {
+		req.Header.Set("Authorization", "Bearer "+key)
+	} else {
+		keyParts := strings.Split(key, "|")
+		if len(keyParts) != 2 {
+			return nil, fmt.Errorf("invalid api key format for jimeng: expected 'ak|sk'")
+		}
+		accessKey := strings.TrimSpace(keyParts[0])
+		secretKey := strings.TrimSpace(keyParts[1])
 
-	if err := a.signRequest(req, accessKey, secretKey); err != nil {
-		return nil, errors.Wrap(err, "sign request failed")
+		if err := a.signRequest(req, accessKey, secretKey); err != nil {
+			return nil, errors.Wrap(err, "sign request failed")
+		}
 	}
-
 	return service.GetHttpClient().Do(req)
 }
 
@@ -384,3 +398,7 @@ func (a *TaskAdaptor) ParseTaskResult(respBody []byte) (*relaycommon.TaskInfo, e
 	taskResult.Url = resTask.Data.VideoUrl
 	return &taskResult, nil
 }
+
+func isNewAPIRelay(apiKey string) bool {
+	return strings.HasPrefix(apiKey, "sk-")
+}

relay/channel/task/kling/adaptor.go

@@ -117,6 +117,11 @@ func (a *TaskAdaptor) ValidateRequestAndSetAction(c *gin.Context, info *relaycom
 // BuildRequestURL constructs the upstream URL.
 func (a *TaskAdaptor) BuildRequestURL(info *relaycommon.RelayInfo) (string, error) {
 	path := lo.Ternary(info.Action == constant.TaskActionGenerate, "/v1/videos/image2video", "/v1/videos/text2video")
+
+	if isNewAPIRelay(info.ApiKey) {
+		return fmt.Sprintf("%s/kling%s", a.baseURL, path), nil
+	}
+
 	return fmt.Sprintf("%s%s", a.baseURL, path), nil
 }
 
@@ -199,6 +204,9 @@ func (a *TaskAdaptor) FetchTask(baseUrl, key string, body map[string]any) (*http
 	}
 	path := lo.Ternary(action == constant.TaskActionGenerate, "/v1/videos/image2video", "/v1/videos/text2video")
 	url := fmt.Sprintf("%s%s/%s", baseUrl, path, taskID)
+	if isNewAPIRelay(key) {
+		url = fmt.Sprintf("%s/kling%s/%s", baseUrl, path, taskID)
+	}
 
 	req, err := http.NewRequest(http.MethodGet, url, nil)
 	if err != nil {
@@ -304,8 +312,13 @@ func (a *TaskAdaptor) createJWTToken() (string, error) {
 //}
 
 func (a *TaskAdaptor) createJWTTokenWithKey(apiKey string) (string, error) {
-
+	if isNewAPIRelay(apiKey) {
+		return apiKey, nil // new api relay
+	}
 	keyParts := strings.Split(apiKey, "|")
+	if len(keyParts) != 2 {
+		return "", errors.New("invalid api_key, required format is accessKey|secretKey")
+	}
 	accessKey := strings.TrimSpace(keyParts[0])
 	if len(keyParts) == 1 {
 		return accessKey, nil
@@ -352,3 +365,7 @@ func (a *TaskAdaptor) ParseTaskResult(respBody []byte) (*relaycommon.TaskInfo, e
 	}
 	return taskInfo, nil
 }
+
+func isNewAPIRelay(apiKey string) bool {
+	return strings.HasPrefix(apiKey, "sk-")
+}

relay/channel/task/vidu/adaptor.go

@@ -80,8 +80,7 @@ func (a *TaskAdaptor) Init(info *relaycommon.RelayInfo) {
 }
 
 func (a *TaskAdaptor) ValidateRequestAndSetAction(c *gin.Context, info *relaycommon.RelayInfo) *dto.TaskError {
-	// Use the unified validation method for TaskSubmitReq with image-based action determination
-	return relaycommon.ValidateTaskRequestWithImageBinding(c, info)
+	return relaycommon.ValidateBasicTaskRequest(c, info, constant.TaskActionGenerate)
 }
 
 func (a *TaskAdaptor) BuildRequestBody(c *gin.Context, _ *relaycommon.RelayInfo) (io.Reader, error) {
@@ -112,6 +111,10 @@ func (a *TaskAdaptor) BuildRequestURL(info *relaycommon.RelayInfo) (string, erro
 	switch info.Action {
 	case constant.TaskActionGenerate:
 		path = "/img2video"
+	case constant.TaskActionFirstTailGenerate:
+		path = "/start-end2video"
+	case constant.TaskActionReferenceGenerate:
+		path = "/reference2video"
 	default:
 		path = "/text2video"
 	}
@@ -187,14 +190,9 @@ func (a *TaskAdaptor) GetChannelName() string {
 // ============================
 
 func (a *TaskAdaptor) convertToRequestPayload(req *relaycommon.TaskSubmitReq) (*requestPayload, error) {
-	var images []string
-	if req.Image != "" {
-		images = []string{req.Image}
-	}
-
 	r := requestPayload{
 		Model:             defaultString(req.Model, "viduq1"),
-		Images:            images,
+		Images:            req.Images,
 		Prompt:            req.Prompt,
 		Duration:          defaultInt(req.Duration, 5),
 		Resolution:        defaultString(req.Size, "1080p"),

relay/common/relay_utils.go

@@ -79,34 +79,18 @@ func ValidateBasicTaskRequest(c *gin.Context, info *RelayInfo, action string) *d
 		req.Images = []string{req.Image}
 	}
 
+	if req.HasImage() {
+		action = constant.TaskActionGenerate
+		if info.ChannelType == constant.ChannelTypeVidu {
+			// vidu 增加 首尾帧生视频和参考图生视频
+			if len(req.Images) == 2 {
+				action = constant.TaskActionFirstTailGenerate
+			} else if len(req.Images) > 2 {
+				action = constant.TaskActionReferenceGenerate
+			}
+		}
+	}
+
 	storeTaskRequest(c, info, action, req)
 	return nil
 }
-
-func ValidateTaskRequestWithImage(c *gin.Context, info *RelayInfo, requestObj interface{}) *dto.TaskError {
-	hasPrompt, ok := requestObj.(HasPrompt)
-	if !ok {
-		return createTaskError(fmt.Errorf("request must have prompt"), "invalid_request", http.StatusBadRequest, true)
-	}
-
-	if taskErr := validatePrompt(hasPrompt.GetPrompt()); taskErr != nil {
-		return taskErr
-	}
-
-	action := constant.TaskActionTextGenerate
-	if hasImage, ok := requestObj.(HasImage); ok && hasImage.HasImage() {
-		action = constant.TaskActionGenerate
-	}
-
-	storeTaskRequest(c, info, action, requestObj)
-	return nil
-}
-
-func ValidateTaskRequestWithImageBinding(c *gin.Context, info *RelayInfo) *dto.TaskError {
-	var req TaskSubmitReq
-	if err := c.ShouldBindJSON(&req); err != nil {
-		return createTaskError(err, "invalid_request_body", http.StatusBadRequest, false)
-	}
-
-	return ValidateTaskRequestWithImage(c, info, req)
-}

relay/compatible_handler.go

@@ -90,41 +90,43 @@ func TextHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *types
 
 		if info.ChannelSetting.SystemPrompt != "" {
 			// 如果有系统提示，则将其添加到请求中
-			request := convertedRequest.(*dto.GeneralOpenAIRequest)
-			containSystemPrompt := false
-			for _, message := range request.Messages {
-				if message.Role == request.GetSystemRoleName() {
-					containSystemPrompt = true
-					break
-				}
-			}
-			if !containSystemPrompt {
-				// 如果没有系统提示，则添加系统提示
-				systemMessage := dto.Message{
-					Role:    request.GetSystemRoleName(),
-					Content: info.ChannelSetting.SystemPrompt,
-				}
-				request.Messages = append([]dto.Message{systemMessage}, request.Messages...)
-			} else if info.ChannelSetting.SystemPromptOverride {
-				common.SetContextKey(c, constant.ContextKeySystemPromptOverride, true)
-				// 如果有系统提示，且允许覆盖，则拼接到前面
-				for i, message := range request.Messages {
+			request, ok := convertedRequest.(*dto.GeneralOpenAIRequest)
+			if ok {
+				containSystemPrompt := false
+				for _, message := range request.Messages {
 					if message.Role == request.GetSystemRoleName() {
-						if message.IsStringContent() {
-							request.Messages[i].SetStringContent(info.ChannelSetting.SystemPrompt + "\n" + message.StringContent())
-						} else {
-							contents := message.ParseContent()
-							contents = append([]dto.MediaContent{
-								{
-									Type: dto.ContentTypeText,
-									Text: info.ChannelSetting.SystemPrompt,
-								},
-							}, contents...)
-							request.Messages[i].Content = contents
-						}
+						containSystemPrompt = true
 						break
 					}
 				}
+				if !containSystemPrompt {
+					// 如果没有系统提示，则添加系统提示
+					systemMessage := dto.Message{
+						Role:    request.GetSystemRoleName(),
+						Content: info.ChannelSetting.SystemPrompt,
+					}
+					request.Messages = append([]dto.Message{systemMessage}, request.Messages...)
+				} else if info.ChannelSetting.SystemPromptOverride {
+					common.SetContextKey(c, constant.ContextKeySystemPromptOverride, true)
+					// 如果有系统提示，且允许覆盖，则拼接到前面
+					for i, message := range request.Messages {
+						if message.Role == request.GetSystemRoleName() {
+							if message.IsStringContent() {
+								request.Messages[i].SetStringContent(info.ChannelSetting.SystemPrompt + "\n" + message.StringContent())
+							} else {
+								contents := message.ParseContent()
+								contents = append([]dto.MediaContent{
+									{
+										Type: dto.ContentTypeText,
+										Text: info.ChannelSetting.SystemPrompt,
+									},
+								}, contents...)
+								request.Messages[i].Content = contents
+							}
+							break
+						}
+					}
+				}
 			}
 		}
 

relay/helper/price.go

@@ -52,6 +52,8 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
 	var cacheRatio float64
 	var imageRatio float64
 	var cacheCreationRatio float64
+	var audioRatio float64
+	var audioCompletionRatio float64
 	if !usePrice {
 		preConsumedTokens := common.Max(promptTokens, common.PreConsumedQuota)
 		if meta.MaxTokens != 0 {
@@ -73,6 +75,8 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
 		cacheRatio, _ = ratio_setting.GetCacheRatio(info.OriginModelName)
 		cacheCreationRatio, _ = ratio_setting.GetCreateCacheRatio(info.OriginModelName)
 		imageRatio, _ = ratio_setting.GetImageRatio(info.OriginModelName)
+		audioRatio = ratio_setting.GetAudioRatio(info.OriginModelName)
+		audioCompletionRatio = ratio_setting.GetAudioCompletionRatio(info.OriginModelName)
 		ratio := modelRatio * groupRatioInfo.GroupRatio
 		preConsumedQuota = int(float64(preConsumedTokens) * ratio)
 	} else {
@@ -90,6 +94,8 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
 		UsePrice:               usePrice,
 		CacheRatio:             cacheRatio,
 		ImageRatio:             imageRatio,
+		AudioRatio:             audioRatio,
+		AudioCompletionRatio:   audioCompletionRatio,
 		CacheCreationRatio:     cacheCreationRatio,
 		ShouldPreConsumedQuota: preConsumedQuota,
 	}

relay/helper/valid_request.go

@@ -21,7 +21,11 @@ func GetAndValidateRequest(c *gin.Context, format types.RelayFormat) (request dt
 	case types.RelayFormatOpenAI:
 		request, err = GetAndValidateTextRequest(c, relayMode)
 	case types.RelayFormatGemini:
-		request, err = GetAndValidateGeminiRequest(c)
+		if strings.Contains(c.Request.URL.Path, ":embedContent") || strings.Contains(c.Request.URL.Path, ":batchEmbedContents") {
+			request, err = GetAndValidateGeminiEmbeddingRequest(c)
+		} else {
+			request, err = GetAndValidateGeminiRequest(c)
+		}
 	case types.RelayFormatClaude:
 		request, err = GetAndValidateClaudeRequest(c)
 	case types.RelayFormatOpenAIResponses:
@@ -288,7 +292,6 @@ func GetAndValidateTextRequest(c *gin.Context, relayMode int) (*dto.GeneralOpenA
 }
 
 func GetAndValidateGeminiRequest(c *gin.Context) (*dto.GeminiChatRequest, error) {
-
 	request := &dto.GeminiChatRequest{}
 	err := common.UnmarshalBodyReusable(c, request)
 	if err != nil {
@@ -304,3 +307,12 @@ func GetAndValidateGeminiRequest(c *gin.Context) (*dto.GeminiChatRequest, error)
 
 	return request, nil
 }
+
+func GetAndValidateGeminiEmbeddingRequest(c *gin.Context) (*dto.GeminiEmbeddingRequest, error) {
+	request := &dto.GeminiEmbeddingRequest{}
+	err := common.UnmarshalBodyReusable(c, request)
+	if err != nil {
+		return nil, err
+	}
+	return request, nil
+}

service/pre_consume_quota.go

@@ -19,7 +19,7 @@ func ReturnPreConsumedQuota(c *gin.Context, relayInfo *relaycommon.RelayInfo) {
 		gopool.Go(func() {
 			relayInfoCopy := *relayInfo
 
-			err := PostConsumeQuota(&relayInfoCopy, -relayInfo.FinalPreConsumedQuota, 0, false)
+			err := PostConsumeQuota(&relayInfoCopy, -relayInfoCopy.FinalPreConsumedQuota, 0, false)
 			if err != nil {
 				common.SysLog("error return pre-consumed quota: " + err.Error())
 			}

setting/ratio_setting/model_ratio.go

@@ -178,6 +178,7 @@ var defaultModelRatio = map[string]float64{
 	"gemini-2.5-flash-lite-preview-thinking-*":  0.05,
 	"gemini-2.5-flash-lite-preview-06-17":       0.05,
 	"gemini-2.5-flash":                          0.15,
+	"gemini-embedding-001":                      0.075,
 	"text-embedding-004":                        0.001,
 	"chatglm_turbo":                             0.3572,     // ￥0.005 / 1k tokens
 	"chatglm_pro":                               0.7143,     // ￥0.01 / 1k tokens
@@ -278,6 +279,18 @@ var defaultModelPrice = map[string]float64{
 	"mj_upload":               0.05,
 }
 
+var defaultAudioRatio = map[string]float64{
+	"gpt-4o-audio-preview":         16,
+	"gpt-4o-mini-audio-preview":    66.67,
+	"gpt-4o-realtime-preview":      8,
+	"gpt-4o-mini-realtime-preview": 16.67,
+}
+
+var defaultAudioCompletionRatio = map[string]float64{
+	"gpt-4o-realtime":      2,
+	"gpt-4o-mini-realtime": 2,
+}
+
 var (
 	modelPriceMap      map[string]float64 = nil
 	modelPriceMapMutex                    = sync.RWMutex{}
@@ -326,6 +339,15 @@ func InitRatioSettings() {
 	imageRatioMap = defaultImageRatio
 	imageRatioMapMutex.Unlock()
 
+	// initialize audioRatioMap
+	audioRatioMapMutex.Lock()
+	audioRatioMap = defaultAudioRatio
+	audioRatioMapMutex.Unlock()
+
+	// initialize audioCompletionRatioMap
+	audioCompletionRatioMapMutex.Lock()
+	audioCompletionRatioMap = defaultAudioCompletionRatio
+	audioCompletionRatioMapMutex.Unlock()
 }
 
 func GetModelPriceMap() map[string]float64 {
@@ -417,6 +439,18 @@ func GetDefaultModelRatioMap() map[string]float64 {
 	return defaultModelRatio
 }
 
+func GetDefaultImageRatioMap() map[string]float64 {
+	return defaultImageRatio
+}
+
+func GetDefaultAudioRatioMap() map[string]float64 {
+	return defaultAudioRatio
+}
+
+func GetDefaultAudioCompletionRatioMap() map[string]float64 {
+	return defaultAudioCompletionRatio
+}
+
 func GetCompletionRatioMap() map[string]float64 {
 	CompletionRatioMutex.RLock()
 	defer CompletionRatioMutex.RUnlock()
@@ -584,32 +618,22 @@ func getHardcodedCompletionModelRatio(name string) (float64, bool) {
 }
 
 func GetAudioRatio(name string) float64 {
-	if strings.Contains(name, "-realtime") {
-		if strings.HasSuffix(name, "gpt-4o-realtime-preview") {
-			return 8
-		} else if strings.Contains(name, "gpt-4o-mini-realtime-preview") {
-			return 10 / 0.6
-		} else {
-			return 20
-		}
-	}
-	if strings.Contains(name, "-audio") {
-		if strings.HasPrefix(name, "gpt-4o-audio-preview") {
-			return 40 / 2.5
-		} else if strings.HasPrefix(name, "gpt-4o-mini-audio-preview") {
-			return 10 / 0.15
-		} else {
-			return 40
-		}
+	audioRatioMapMutex.RLock()
+	defer audioRatioMapMutex.RUnlock()
+	name = FormatMatchingModelName(name)
+	if ratio, ok := audioRatioMap[name]; ok {
+		return ratio
 	}
 	return 20
 }
 
 func GetAudioCompletionRatio(name string) float64 {
-	if strings.HasPrefix(name, "gpt-4o-realtime") {
-		return 2
-	} else if strings.HasPrefix(name, "gpt-4o-mini-realtime") {
-		return 2
+	audioCompletionRatioMapMutex.RLock()
+	defer audioCompletionRatioMapMutex.RUnlock()
+	name = FormatMatchingModelName(name)
+	if ratio, ok := audioCompletionRatioMap[name]; ok {
+
+		return ratio
 	}
 	return 2
 }
@@ -630,6 +654,14 @@ var defaultImageRatio = map[string]float64{
 }
 var imageRatioMap map[string]float64
 var imageRatioMapMutex sync.RWMutex
+var (
+	audioRatioMap      map[string]float64 = nil
+	audioRatioMapMutex                    = sync.RWMutex{}
+)
+var (
+	audioCompletionRatioMap      map[string]float64 = nil
+	audioCompletionRatioMapMutex                    = sync.RWMutex{}
+)
 
 func ImageRatio2JSONString() string {
 	imageRatioMapMutex.RLock()
@@ -658,6 +690,71 @@ func GetImageRatio(name string) (float64, bool) {
 	return ratio, true
 }
 
+func AudioRatio2JSONString() string {
+	audioRatioMapMutex.RLock()
+	defer audioRatioMapMutex.RUnlock()
+	jsonBytes, err := common.Marshal(audioRatioMap)
+	if err != nil {
+		common.SysError("error marshalling audio ratio: " + err.Error())
+	}
+	return string(jsonBytes)
+}
+
+func UpdateAudioRatioByJSONString(jsonStr string) error {
+
+	tmp := make(map[string]float64)
+	if err := common.Unmarshal([]byte(jsonStr), &tmp); err != nil {
+		return err
+	}
+	audioRatioMapMutex.Lock()
+	audioRatioMap = tmp
+	audioRatioMapMutex.Unlock()
+	InvalidateExposedDataCache()
+	return nil
+}
+
+func GetAudioRatioCopy() map[string]float64 {
+	audioRatioMapMutex.RLock()
+	defer audioRatioMapMutex.RUnlock()
+	copyMap := make(map[string]float64, len(audioRatioMap))
+	for k, v := range audioRatioMap {
+		copyMap[k] = v
+	}
+	return copyMap
+}
+
+func AudioCompletionRatio2JSONString() string {
+	audioCompletionRatioMapMutex.RLock()
+	defer audioCompletionRatioMapMutex.RUnlock()
+	jsonBytes, err := common.Marshal(audioCompletionRatioMap)
+	if err != nil {
+		common.SysError("error marshalling audio completion ratio: " + err.Error())
+	}
+	return string(jsonBytes)
+}
+
+func UpdateAudioCompletionRatioByJSONString(jsonStr string) error {
+	tmp := make(map[string]float64)
+	if err := common.Unmarshal([]byte(jsonStr), &tmp); err != nil {
+		return err
+	}
+	audioCompletionRatioMapMutex.Lock()
+	audioCompletionRatioMap = tmp
+	audioCompletionRatioMapMutex.Unlock()
+	InvalidateExposedDataCache()
+	return nil
+}
+
+func GetAudioCompletionRatioCopy() map[string]float64 {
+	audioCompletionRatioMapMutex.RLock()
+	defer audioCompletionRatioMapMutex.RUnlock()
+	copyMap := make(map[string]float64, len(audioCompletionRatioMap))
+	for k, v := range audioCompletionRatioMap {
+		copyMap[k] = v
+	}
+	return copyMap
+}
+
 func GetModelRatioCopy() map[string]float64 {
 	modelRatioMapMutex.RLock()
 	defer modelRatioMapMutex.RUnlock()

types/price_data.go

@@ -15,6 +15,8 @@ type PriceData struct {
 	CacheRatio             float64
 	CacheCreationRatio     float64
 	ImageRatio             float64
+	AudioRatio             float64
+	AudioCompletionRatio   float64
 	UsePrice               bool
 	ShouldPreConsumedQuota int
 	GroupRatioInfo         GroupRatioInfo
@@ -27,5 +29,5 @@ type PerCallPriceData struct {
 }
 
 func (p PriceData) ToSetting() string {
-	return fmt.Sprintf("ModelPrice: %f, ModelRatio: %f, CompletionRatio: %f, CacheRatio: %f, GroupRatio: %f, UsePrice: %t, CacheCreationRatio: %f, ShouldPreConsumedQuota: %d, ImageRatio: %f", p.ModelPrice, p.ModelRatio, p.CompletionRatio, p.CacheRatio, p.GroupRatioInfo.GroupRatio, p.UsePrice, p.CacheCreationRatio, p.ShouldPreConsumedQuota, p.ImageRatio)
+	return fmt.Sprintf("ModelPrice: %f, ModelRatio: %f, CompletionRatio: %f, CacheRatio: %f, GroupRatio: %f, UsePrice: %t, CacheCreationRatio: %f, ShouldPreConsumedQuota: %d, ImageRatio: %f, AudioRatio: %f, AudioCompletionRatio: %f", p.ModelPrice, p.ModelRatio, p.CompletionRatio, p.CacheRatio, p.GroupRatioInfo.GroupRatio, p.UsePrice, p.CacheCreationRatio, p.ShouldPreConsumedQuota, p.ImageRatio, p.AudioRatio, p.AudioCompletionRatio)
 }

web/src/components/settings/RatioSetting.jsx

@@ -39,6 +39,9 @@ const RatioSetting = () => {
     CompletionRatio: '',
     GroupRatio: '',
     GroupGroupRatio: '',
+    ImageRatio: '',
+    AudioRatio: '',
+    AudioCompletionRatio: '',
     AutoGroups: '',
     DefaultUseAutoGroup: false,
     ExposeRatioEnabled: false,
@@ -61,7 +64,10 @@ const RatioSetting = () => {
           item.key === 'UserUsableGroups' ||
           item.key === 'CompletionRatio' ||
           item.key === 'ModelPrice' ||
-          item.key === 'CacheRatio'
+          item.key === 'CacheRatio' ||
+          item.key === 'ImageRatio' ||
+          item.key === 'AudioRatio' ||
+          item.key === 'AudioCompletionRatio'
         ) {
           try {
             item.value = JSON.stringify(JSON.parse(item.value), null, 2);

web/src/components/table/task-logs/TaskLogsColumnDefs.jsx

@@ -35,8 +35,9 @@ import {
   Sparkles,
 } from 'lucide-react';
 import {
-  TASK_ACTION_GENERATE,
-  TASK_ACTION_TEXT_GENERATE,
+  TASK_ACTION_FIRST_TAIL_GENERATE,
+  TASK_ACTION_GENERATE, TASK_ACTION_REFERENCE_GENERATE,
+  TASK_ACTION_TEXT_GENERATE
 } from '../../../constants/common.constant';
 import { CHANNEL_OPTIONS } from '../../../constants/channel.constants';
 
@@ -111,6 +112,18 @@ const renderType = (type, t) => {
           {t('文生视频')}
         </Tag>
       );
+    case TASK_ACTION_FIRST_TAIL_GENERATE:
+      return (
+        <Tag color='blue' shape='circle' prefixIcon={<Sparkles size={14} />}>
+          {t('首尾生视频')}
+        </Tag>
+      );
+    case TASK_ACTION_REFERENCE_GENERATE:
+      return (
+        <Tag color='blue' shape='circle' prefixIcon={<Sparkles size={14} />}>
+          {t('参照生视频')}
+        </Tag>
+      );
     default:
       return (
         <Tag color='white' shape='circle' prefixIcon={<HelpCircle size={14} />}>
@@ -343,7 +356,9 @@ export const getTaskLogsColumns = ({
         // 仅当为视频生成任务且成功，且 fail_reason 是 URL 时显示可点击链接
         const isVideoTask =
           record.action === TASK_ACTION_GENERATE ||
-          record.action === TASK_ACTION_TEXT_GENERATE;
+          record.action === TASK_ACTION_TEXT_GENERATE ||
+          record.action === TASK_ACTION_FIRST_TAIL_GENERATE ||
+          record.action === TASK_ACTION_REFERENCE_GENERATE;
         const isSuccess = record.status === 'SUCCESS';
         const isUrl = typeof text === 'string' && /^https?:\/\//.test(text);
         if (isSuccess && isVideoTask && isUrl) {

web/src/constants/common.constant.js

@@ -40,3 +40,5 @@ export const API_ENDPOINTS = [
 
 export const TASK_ACTION_GENERATE = 'generate';
 export const TASK_ACTION_TEXT_GENERATE = 'textGenerate';
+export const TASK_ACTION_FIRST_TAIL_GENERATE = 'firstTailGenerate';
+export const TASK_ACTION_REFERENCE_GENERATE = 'referenceGenerate';

web/src/i18n/locales/en.json

@@ -1999,6 +1999,16 @@
   "查看渠道密钥": "View channel key",
   "渠道密钥信息": "Channel key information",
   "密钥获取成功": "Key acquisition successful",
+  "模型补全倍率（仅对自定义模型有效）": "Model completion ratio (only effective for custom models)",
+  "图片倍率": "Image ratio",
+  "音频倍率": "Audio ratio",
+  "音频补全倍率": "Audio completion ratio",
+  "图片输入相关的倍率设置，键为模型名称，值为倍率": "Image input related ratio settings, key is model name, value is ratio",
+  "音频输入相关的倍率设置，键为模型名称，值为倍率": "Audio input related ratio settings, key is model name, value is ratio",
+  "音频输出补全相关的倍率设置，键为模型名称，值为倍率": "Audio output completion related ratio settings, key is model name, value is ratio",
+  "为一个 JSON 文本，键为模型名称，值为倍率，例如：{\"gpt-image-1\": 2}": "A JSON text with model name as key and ratio as value, e.g.: {\"gpt-image-1\": 2}",
+  "为一个 JSON 文本，键为模型名称，值为倍率，例如：{\"gpt-4o-audio-preview\": 16}": "A JSON text with model name as key and ratio as value, e.g.: {\"gpt-4o-audio-preview\": 16}",
+  "为一个 JSON 文本，键为模型名称，值为倍率，例如：{\"gpt-4o-realtime\": 2}": "A JSON text with model name as key and ratio as value, e.g.: {\"gpt-4o-realtime\": 2}",
   "顶栏管理": "Header Management",
   "控制顶栏模块显示状态，全局生效": "Control header module display status, global effect",
   "用户主页，展示系统信息": "User homepage, displaying system information",
@@ -2058,7 +2068,7 @@
   "需要登录访问": "Require Login",
   "开启后未登录用户无法访问模型广场": "When enabled, unauthenticated users cannot access the model marketplace",
   "参与官方同步": "Participate in official sync",
-  "关闭后，此模型将不会被“同步官方”自动覆盖或创建": "When turned off, this model will be skipped by Sync official (no auto create/overwrite)",
+  "关闭后，此模型将不会被\"同步官方\"自动覆盖或创建": "When turned off, this model will be skipped by Sync official (no auto create/overwrite)",
   "同步": "Sync",
   "同步向导": "Sync Wizard",
   "选择方式": "Select method",

web/src/pages/Setting/Ratio/ModelRatioSettings.jsx

@@ -44,6 +44,9 @@ export default function ModelRatioSettings(props) {
     ModelRatio: '',
     CacheRatio: '',
     CompletionRatio: '',
+    ImageRatio: '',
+    AudioRatio: '',
+    AudioCompletionRatio: '',
     ExposeRatioEnabled: false,
   });
   const refForm = useRef();
@@ -219,6 +222,72 @@ export default function ModelRatioSettings(props) {
             />
           </Col>
         </Row>
+        <Row gutter={16}>
+          <Col xs={24} sm={16}>
+            <Form.TextArea
+              label={t('图片输入倍率（仅部分模型支持该计费）')}
+              extraText={t('图片输入相关的倍率设置，键为模型名称，值为倍率，仅部分模型支持该计费')}
+              placeholder={t('为一个 JSON 文本，键为模型名称，值为倍率，例如：{"gpt-image-1": 2}')}
+              field={'ImageRatio'}
+              autosize={{ minRows: 6, maxRows: 12 }}
+              trigger='blur'
+              stopValidateWithError
+              rules={[
+                {
+                  validator: (rule, value) => verifyJSON(value),
+                  message: '不是合法的 JSON 字符串',
+                },
+              ]}
+              onChange={(value) =>
+                setInputs({ ...inputs, ImageRatio: value })
+              }
+            />
+          </Col>
+        </Row>
+        <Row gutter={16}>
+          <Col xs={24} sm={16}>
+            <Form.TextArea
+              label={t('音频倍率（仅部分模型支持该计费）')}
+              extraText={t('音频输入相关的倍率设置，键为模型名称，值为倍率')}
+              placeholder={t('为一个 JSON 文本，键为模型名称，值为倍率，例如：{"gpt-4o-audio-preview": 16}')}
+              field={'AudioRatio'}
+              autosize={{ minRows: 6, maxRows: 12 }}
+              trigger='blur'
+              stopValidateWithError
+              rules={[
+                {
+                  validator: (rule, value) => verifyJSON(value),
+                  message: '不是合法的 JSON 字符串',
+                },
+              ]}
+              onChange={(value) =>
+                setInputs({ ...inputs, AudioRatio: value })
+              }
+            />
+          </Col>
+        </Row>
+        <Row gutter={16}>
+          <Col xs={24} sm={16}>
+            <Form.TextArea
+              label={t('音频补全倍率（仅部分模型支持该计费）')}
+              extraText={t('音频输出补全相关的倍率设置，键为模型名称，值为倍率')}
+              placeholder={t('为一个 JSON 文本，键为模型名称，值为倍率，例如：{"gpt-4o-realtime": 2}')}
+              field={'AudioCompletionRatio'}
+              autosize={{ minRows: 6, maxRows: 12 }}
+              trigger='blur'
+              stopValidateWithError
+              rules={[
+                {
+                  validator: (rule, value) => verifyJSON(value),
+                  message: '不是合法的 JSON 字符串',
+                },
+              ]}
+              onChange={(value) =>
+                setInputs({ ...inputs, AudioCompletionRatio: value })
+              }
+            />
+          </Col>
+        </Row>
         <Row gutter={16}>
           <Col span={16}>
             <Form.Switch