diff --git a/dto/dalle.go b/dto/dalle.go
index d0bba655..562d5f1a 100644
--- a/dto/dalle.go
+++ b/dto/dalle.go
@@ -1,14 +1,17 @@
package dto
+import "encoding/json"
+
type ImageRequest struct {
- Model string `json:"model"`
- Prompt string `json:"prompt" binding:"required"`
- N int `json:"n,omitempty"`
- Size string `json:"size,omitempty"`
- Quality string `json:"quality,omitempty"`
- ResponseFormat string `json:"response_format,omitempty"`
- Style string `json:"style,omitempty"`
- User string `json:"user,omitempty"`
+ Model string `json:"model"`
+ Prompt string `json:"prompt" binding:"required"`
+ N int `json:"n,omitempty"`
+ Size string `json:"size,omitempty"`
+ Quality string `json:"quality,omitempty"`
+ ResponseFormat string `json:"response_format,omitempty"`
+ Style string `json:"style,omitempty"`
+ User string `json:"user,omitempty"`
+ ExtraFields json.RawMessage `json:"extra_fields,omitempty"`
}
type ImageResponse struct {
diff --git a/dto/openai_response.go b/dto/openai_response.go
index fe2609bf..ddd1a907 100644
--- a/dto/openai_response.go
+++ b/dto/openai_response.go
@@ -173,3 +173,17 @@ type Usage struct {
PromptTokensDetails InputTokenDetails `json:"prompt_tokens_details"`
CompletionTokenDetails OutputTokenDetails `json:"completion_tokens_details"`
}
+
+type InputTokenDetails struct {
+ CachedTokens int `json:"cached_tokens"`
+ CachedCreationTokens int `json:"-"`
+ TextTokens int `json:"text_tokens"`
+ AudioTokens int `json:"audio_tokens"`
+ ImageTokens int `json:"image_tokens"`
+}
+
+type OutputTokenDetails struct {
+ TextTokens int `json:"text_tokens"`
+ AudioTokens int `json:"audio_tokens"`
+ ReasoningTokens int `json:"reasoning_tokens"`
+}
diff --git a/dto/realtime.go b/dto/realtime.go
index bb572267..86ae352d 100644
--- a/dto/realtime.go
+++ b/dto/realtime.go
@@ -43,20 +43,6 @@ type RealtimeUsage struct {
OutputTokenDetails OutputTokenDetails `json:"output_token_details"`
}
-type InputTokenDetails struct {
- CachedTokens int `json:"cached_tokens"`
- CachedCreationTokens int `json:"-"`
- TextTokens int `json:"text_tokens"`
- AudioTokens int `json:"audio_tokens"`
- ImageTokens int `json:"image_tokens"`
-}
-
-type OutputTokenDetails struct {
- TextTokens int `json:"text_tokens"`
- AudioTokens int `json:"audio_tokens"`
- ReasoningTokens int `json:"reasoning_tokens"`
-}
-
type RealtimeSession struct {
Modalities []string `json:"modalities"`
Instructions string `json:"instructions"`
diff --git a/relay/channel/gemini/adaptor.go b/relay/channel/gemini/adaptor.go
index e5ee134a..feaed8f4 100644
--- a/relay/channel/gemini/adaptor.go
+++ b/relay/channel/gemini/adaptor.go
@@ -12,7 +12,6 @@ import (
relaycommon "one-api/relay/common"
"one-api/service"
"one-api/setting/model_setting"
-
"strings"
"github.com/gin-gonic/gin"
@@ -70,6 +69,16 @@ func (a *Adaptor) Init(info *relaycommon.RelayInfo) {
}
func (a *Adaptor) GetRequestURL(info *relaycommon.RelayInfo) (string, error) {
+
+ if model_setting.GetGeminiSettings().ThinkingAdapterEnabled {
+ // suffix -thinking and -nothinking
+ if strings.HasSuffix(info.OriginModelName, "-thinking") {
+ info.UpstreamModelName = strings.TrimSuffix(info.UpstreamModelName, "-thinking")
+ } else if strings.HasSuffix(info.OriginModelName, "-nothinking") {
+ info.UpstreamModelName = strings.TrimSuffix(info.UpstreamModelName, "-nothinking")
+ }
+ }
+
version := model_setting.GetGeminiVersionSetting(info.UpstreamModelName)
if strings.HasPrefix(info.UpstreamModelName, "imagen") {
@@ -99,11 +108,13 @@ func (a *Adaptor) ConvertOpenAIRequest(c *gin.Context, info *relaycommon.RelayIn
if request == nil {
return nil, errors.New("request is nil")
}
- ai, err := CovertGemini2OpenAI(*request, info)
+
+ geminiRequest, err := CovertGemini2OpenAI(*request, info)
if err != nil {
return nil, err
}
- return ai, nil
+
+ return geminiRequest, nil
}
func (a *Adaptor) ConvertRerankRequest(c *gin.Context, relayMode int, request dto.RerankRequest) (any, error) {
@@ -165,6 +176,18 @@ func (a *Adaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycom
} else {
err, usage = GeminiChatHandler(c, resp, info)
}
+
+ //if usage.(*dto.Usage).CompletionTokenDetails.ReasoningTokens > 100 {
+ // // 没有请求-thinking的情况下，产生思考token，则按照思考模型计费
+ // if !strings.HasSuffix(info.OriginModelName, "-thinking") &&
+ // !strings.HasSuffix(info.OriginModelName, "-nothinking") {
+ // thinkingModelName := info.OriginModelName + "-thinking"
+ // if operation_setting.SelfUseModeEnabled || helper.ContainPriceOrRatio(thinkingModelName) {
+ // info.OriginModelName = thinkingModelName
+ // }
+ // }
+ //}
+
return
}
diff --git a/relay/channel/gemini/dto.go b/relay/channel/gemini/dto.go
index 7f98b1b7..5d5c1287 100644
--- a/relay/channel/gemini/dto.go
+++ b/relay/channel/gemini/dto.go
@@ -8,6 +8,15 @@ type GeminiChatRequest struct {
SystemInstructions *GeminiChatContent `json:"system_instruction,omitempty"`
}
+type GeminiThinkingConfig struct {
+ IncludeThoughts bool `json:"includeThoughts,omitempty"`
+ ThinkingBudget *int `json:"thinkingBudget,omitempty"`
+}
+
+func (c *GeminiThinkingConfig) SetThinkingBudget(budget int) {
+ c.ThinkingBudget = &budget
+}
+
type GeminiInlineData struct {
MimeType string `json:"mimeType"`
Data string `json:"data"`
@@ -71,16 +80,17 @@ type GeminiChatTool struct {
}
type GeminiChatGenerationConfig struct {
- Temperature *float64 `json:"temperature,omitempty"`
- TopP float64 `json:"topP,omitempty"`
- TopK float64 `json:"topK,omitempty"`
- MaxOutputTokens uint `json:"maxOutputTokens,omitempty"`
- CandidateCount int `json:"candidateCount,omitempty"`
- StopSequences []string `json:"stopSequences,omitempty"`
- ResponseMimeType string `json:"responseMimeType,omitempty"`
- ResponseSchema any `json:"responseSchema,omitempty"`
- Seed int64 `json:"seed,omitempty"`
- ResponseModalities []string `json:"responseModalities,omitempty"`
+ Temperature *float64 `json:"temperature,omitempty"`
+ TopP float64 `json:"topP,omitempty"`
+ TopK float64 `json:"topK,omitempty"`
+ MaxOutputTokens uint `json:"maxOutputTokens,omitempty"`
+ CandidateCount int `json:"candidateCount,omitempty"`
+ StopSequences []string `json:"stopSequences,omitempty"`
+ ResponseMimeType string `json:"responseMimeType,omitempty"`
+ ResponseSchema any `json:"responseSchema,omitempty"`
+ Seed int64 `json:"seed,omitempty"`
+ ResponseModalities []string `json:"responseModalities,omitempty"`
+ ThinkingConfig *GeminiThinkingConfig `json:"thinkingConfig,omitempty"`
}
type GeminiChatCandidate struct {
@@ -109,6 +119,7 @@ type GeminiUsageMetadata struct {
PromptTokenCount int `json:"promptTokenCount"`
CandidatesTokenCount int `json:"candidatesTokenCount"`
TotalTokenCount int `json:"totalTokenCount"`
+ ThoughtsTokenCount int `json:"thoughtsTokenCount"`
}
// Imagen related structs
diff --git a/relay/channel/gemini/relay-gemini.go b/relay/channel/gemini/relay-gemini.go
index 03736f38..e358d741 100644
--- a/relay/channel/gemini/relay-gemini.go
+++ b/relay/channel/gemini/relay-gemini.go
@@ -23,12 +23,14 @@ func CovertGemini2OpenAI(textRequest dto.GeneralOpenAIRequest, info *relaycommon
geminiRequest := GeminiChatRequest{
Contents: make([]GeminiChatContent, 0, len(textRequest.Messages)),
- //SafetySettings: []GeminiChatSafetySettings{},
GenerationConfig: GeminiChatGenerationConfig{
Temperature: textRequest.Temperature,
TopP: textRequest.TopP,
MaxOutputTokens: textRequest.MaxTokens,
Seed: int64(textRequest.Seed),
+ ThinkingConfig: &GeminiThinkingConfig{
+ IncludeThoughts: true,
+ },
},
}
@@ -39,6 +41,18 @@ func CovertGemini2OpenAI(textRequest dto.GeneralOpenAIRequest, info *relaycommon
}
}
+ if model_setting.GetGeminiSettings().ThinkingAdapterEnabled {
+ if strings.HasSuffix(info.OriginModelName, "-thinking") {
+ budgetTokens := model_setting.GetGeminiSettings().ThinkingAdapterBudgetTokensPercentage * float64(geminiRequest.GenerationConfig.MaxOutputTokens)
+ if budgetTokens == 0 || budgetTokens > 24576 {
+ budgetTokens = 24576
+ }
+ geminiRequest.GenerationConfig.ThinkingConfig.SetThinkingBudget(int(budgetTokens))
+ } else if strings.HasSuffix(info.OriginModelName, "-nothinking") {
+ geminiRequest.GenerationConfig.ThinkingConfig.SetThinkingBudget(0)
+ }
+ }
+
safetySettings := make([]GeminiChatSafetySettings, 0, len(SafetySettingList))
for _, category := range SafetySettingList {
safetySettings = append(safetySettings, GeminiChatSafetySettings{
@@ -644,6 +658,7 @@ func GeminiChatStreamHandler(c *gin.Context, resp *http.Response, info *relaycom
if geminiResponse.UsageMetadata.TotalTokenCount != 0 {
usage.PromptTokens = geminiResponse.UsageMetadata.PromptTokenCount
usage.CompletionTokens = geminiResponse.UsageMetadata.CandidatesTokenCount
+ usage.CompletionTokenDetails.ReasoningTokens = geminiResponse.UsageMetadata.ThoughtsTokenCount
}
err = helper.ObjectData(c, response)
if err != nil {
@@ -666,7 +681,7 @@ func GeminiChatStreamHandler(c *gin.Context, resp *http.Response, info *relaycom
usage.TotalTokens = usage.PromptTokens + usage.CompletionTokens
usage.PromptTokensDetails.TextTokens = usage.PromptTokens
- usage.CompletionTokenDetails.TextTokens = usage.CompletionTokens
+ //usage.CompletionTokenDetails.TextTokens = usage.CompletionTokens
if info.ShouldIncludeUsage {
response = helper.GenerateFinalUsageResponse(id, createAt, info.UpstreamModelName, *usage)
@@ -712,6 +727,9 @@ func GeminiChatHandler(c *gin.Context, resp *http.Response, info *relaycommon.Re
CompletionTokens: geminiResponse.UsageMetadata.CandidatesTokenCount,
TotalTokens: geminiResponse.UsageMetadata.TotalTokenCount,
}
+
+ usage.CompletionTokenDetails.ReasoningTokens = geminiResponse.UsageMetadata.ThoughtsTokenCount
+
fullTextResponse.Usage = usage
jsonResponse, err := json.Marshal(fullTextResponse)
if err != nil {
diff --git a/relay/helper/price.go b/relay/helper/price.go
index 758c2829..a68cd54d 100644
--- a/relay/helper/price.go
+++ b/relay/helper/price.go
@@ -49,11 +49,7 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
}
}
if !acceptUnsetRatio {
- if info.UserId == 1 {
- return PriceData{}, fmt.Errorf("模型 %s 倍率或价格未配置，请设置或开始自用模式；Model %s ratio or price not set, please set or start self-use mode", info.OriginModelName, info.OriginModelName)
- } else {
- return PriceData{}, fmt.Errorf("模型 %s 倍率或价格未配置, 请联系管理员设置；Model %s ratio or price not set, please contact administrator to set", info.OriginModelName, info.OriginModelName)
- }
+ return PriceData{}, fmt.Errorf("模型 %s 倍率或价格未配置，请联系管理员设置或开始自用模式；Model %s ratio or price not set, please set or start self-use mode", info.OriginModelName, info.OriginModelName)
}
}
completionRatio = operation_setting.GetCompletionRatio(info.OriginModelName)
@@ -82,3 +78,15 @@ func ModelPriceHelper(c *gin.Context, info *relaycommon.RelayInfo, promptTokens
return priceData, nil
}
+
+func ContainPriceOrRatio(modelName string) bool {
+ _, ok := operation_setting.GetModelPrice(modelName, false)
+ if ok {
+ return true
+ }
+ _, ok = operation_setting.GetModelRatio(modelName)
+ if ok {
+ return true
+ }
+ return false
+}
diff --git a/setting/model_setting/gemini.go b/setting/model_setting/gemini.go
index e6509232..f132fec8 100644
--- a/setting/model_setting/gemini.go
+++ b/setting/model_setting/gemini.go
@@ -6,9 +6,11 @@ import (
// GeminiSettings 定义Gemini模型的配置
type GeminiSettings struct {
- SafetySettings map[string]string `json:"safety_settings"`
- VersionSettings map[string]string `json:"version_settings"`
- SupportedImagineModels []string `json:"supported_imagine_models"`
+ SafetySettings map[string]string `json:"safety_settings"`
+ VersionSettings map[string]string `json:"version_settings"`
+ SupportedImagineModels []string `json:"supported_imagine_models"`
+ ThinkingAdapterEnabled bool `json:"thinking_adapter_enabled"`
+ ThinkingAdapterBudgetTokensPercentage float64 `json:"thinking_adapter_budget_tokens_percentage"`
}
// 默认配置
@@ -25,6 +27,8 @@ var defaultGeminiSettings = GeminiSettings{
"gemini-2.0-flash-exp-image-generation",
"gemini-2.0-flash-exp",
},
+ ThinkingAdapterEnabled: false,
+ ThinkingAdapterBudgetTokensPercentage: 0.6,
}
// 全局实例
diff --git a/setting/operation_setting/model-ratio.go b/setting/operation_setting/model-ratio.go
index 68e50757..6a80ef1a 100644
--- a/setting/operation_setting/model-ratio.go
+++ b/setting/operation_setting/model-ratio.go
@@ -86,89 +86,92 @@ var defaultModelRatio = map[string]float64{
"text-curie-001": 1,
//"text-davinci-002": 10,
//"text-davinci-003": 10,
- "text-davinci-edit-001": 10,
- "code-davinci-edit-001": 10,
- "whisper-1": 15, // $0.006 / minute -> $0.006 / 150 words -> $0.006 / 200 tokens -> $0.03 / 1k tokens
- "tts-1": 7.5, // 1k characters -> $0.015
- "tts-1-1106": 7.5, // 1k characters -> $0.015
- "tts-1-hd": 15, // 1k characters -> $0.03
- "tts-1-hd-1106": 15, // 1k characters -> $0.03
- "davinci": 10,
- "curie": 10,
- "babbage": 10,
- "ada": 10,
- "text-embedding-3-small": 0.01,
- "text-embedding-3-large": 0.065,
- "text-embedding-ada-002": 0.05,
- "text-search-ada-doc-001": 10,
- "text-moderation-stable": 0.1,
- "text-moderation-latest": 0.1,
- "claude-instant-1": 0.4, // $0.8 / 1M tokens
- "claude-2.0": 4, // $8 / 1M tokens
- "claude-2.1": 4, // $8 / 1M tokens
- "claude-3-haiku-20240307": 0.125, // $0.25 / 1M tokens
- "claude-3-5-haiku-20241022": 0.5, // $1 / 1M tokens
- "claude-3-sonnet-20240229": 1.5, // $3 / 1M tokens
- "claude-3-5-sonnet-20240620": 1.5,
- "claude-3-5-sonnet-20241022": 1.5,
- "claude-3-7-sonnet-20250219": 1.5,
- "claude-3-7-sonnet-20250219-thinking": 1.5,
- "claude-3-opus-20240229": 7.5, // $15 / 1M tokens
- "ERNIE-4.0-8K": 0.120 * RMB,
- "ERNIE-3.5-8K": 0.012 * RMB,
- "ERNIE-3.5-8K-0205": 0.024 * RMB,
- "ERNIE-3.5-8K-1222": 0.012 * RMB,
- "ERNIE-Bot-8K": 0.024 * RMB,
- "ERNIE-3.5-4K-0205": 0.012 * RMB,
- "ERNIE-Speed-8K": 0.004 * RMB,
- "ERNIE-Speed-128K": 0.004 * RMB,
- "ERNIE-Lite-8K-0922": 0.008 * RMB,
- "ERNIE-Lite-8K-0308": 0.003 * RMB,
- "ERNIE-Tiny-8K": 0.001 * RMB,
- "BLOOMZ-7B": 0.004 * RMB,
- "Embedding-V1": 0.002 * RMB,
- "bge-large-zh": 0.002 * RMB,
- "bge-large-en": 0.002 * RMB,
- "tao-8k": 0.002 * RMB,
- "PaLM-2": 1,
- "gemini-1.5-pro-latest": 1.25, // $3.5 / 1M tokens
- "gemini-1.5-flash-latest": 0.075,
- "gemini-2.0-flash": 0.05,
- "gemini-2.5-pro-exp-03-25": 0.625,
- "gemini-2.5-pro-preview-03-25": 0.625,
- "text-embedding-004": 0.001,
- "chatglm_turbo": 0.3572, // ￥0.005 / 1k tokens
- "chatglm_pro": 0.7143, // ￥0.01 / 1k tokens
- "chatglm_std": 0.3572, // ￥0.005 / 1k tokens
- "chatglm_lite": 0.1429, // ￥0.002 / 1k tokens
- "glm-4": 7.143, // ￥0.1 / 1k tokens
- "glm-4v": 0.05 * RMB, // ￥0.05 / 1k tokens
- "glm-4-alltools": 0.1 * RMB, // ￥0.1 / 1k tokens
- "glm-3-turbo": 0.3572,
- "glm-4-plus": 0.05 * RMB,
- "glm-4-0520": 0.1 * RMB,
- "glm-4-air": 0.001 * RMB,
- "glm-4-airx": 0.01 * RMB,
- "glm-4-long": 0.001 * RMB,
- "glm-4-flash": 0,
- "glm-4v-plus": 0.01 * RMB,
- "qwen-turbo": 0.8572, // ￥0.012 / 1k tokens
- "qwen-plus": 10, // ￥0.14 / 1k tokens
- "text-embedding-v1": 0.05, // ￥0.0007 / 1k tokens
- "SparkDesk-v1.1": 1.2858, // ￥0.018 / 1k tokens
- "SparkDesk-v2.1": 1.2858, // ￥0.018 / 1k tokens
- "SparkDesk-v3.1": 1.2858, // ￥0.018 / 1k tokens
- "SparkDesk-v3.5": 1.2858, // ￥0.018 / 1k tokens
- "SparkDesk-v4.0": 1.2858,
- "360GPT_S2_V9": 0.8572, // ¥0.012 / 1k tokens
- "360gpt-turbo": 0.0858, // ¥0.0012 / 1k tokens
- "360gpt-turbo-responsibility-8k": 0.8572, // ¥0.012 / 1k tokens
- "360gpt-pro": 0.8572, // ¥0.012 / 1k tokens
- "360gpt2-pro": 0.8572, // ¥0.012 / 1k tokens
- "embedding-bert-512-v1": 0.0715, // ¥0.001 / 1k tokens
- "embedding_s1_v1": 0.0715, // ¥0.001 / 1k tokens
- "semantic_similarity_s1_v1": 0.0715, // ¥0.001 / 1k tokens
- "hunyuan": 7.143, // ¥0.1 / 1k tokens // https://cloud.tencent.com/document/product/1729/97731#e0e6be58-60c8-469f-bdeb-6c264ce3b4d0
+ "text-davinci-edit-001": 10,
+ "code-davinci-edit-001": 10,
+ "whisper-1": 15, // $0.006 / minute -> $0.006 / 150 words -> $0.006 / 200 tokens -> $0.03 / 1k tokens
+ "tts-1": 7.5, // 1k characters -> $0.015
+ "tts-1-1106": 7.5, // 1k characters -> $0.015
+ "tts-1-hd": 15, // 1k characters -> $0.03
+ "tts-1-hd-1106": 15, // 1k characters -> $0.03
+ "davinci": 10,
+ "curie": 10,
+ "babbage": 10,
+ "ada": 10,
+ "text-embedding-3-small": 0.01,
+ "text-embedding-3-large": 0.065,
+ "text-embedding-ada-002": 0.05,
+ "text-search-ada-doc-001": 10,
+ "text-moderation-stable": 0.1,
+ "text-moderation-latest": 0.1,
+ "claude-instant-1": 0.4, // $0.8 / 1M tokens
+ "claude-2.0": 4, // $8 / 1M tokens
+ "claude-2.1": 4, // $8 / 1M tokens
+ "claude-3-haiku-20240307": 0.125, // $0.25 / 1M tokens
+ "claude-3-5-haiku-20241022": 0.5, // $1 / 1M tokens
+ "claude-3-sonnet-20240229": 1.5, // $3 / 1M tokens
+ "claude-3-5-sonnet-20240620": 1.5,
+ "claude-3-5-sonnet-20241022": 1.5,
+ "claude-3-7-sonnet-20250219": 1.5,
+ "claude-3-7-sonnet-20250219-thinking": 1.5,
+ "claude-3-opus-20240229": 7.5, // $15 / 1M tokens
+ "ERNIE-4.0-8K": 0.120 * RMB,
+ "ERNIE-3.5-8K": 0.012 * RMB,
+ "ERNIE-3.5-8K-0205": 0.024 * RMB,
+ "ERNIE-3.5-8K-1222": 0.012 * RMB,
+ "ERNIE-Bot-8K": 0.024 * RMB,
+ "ERNIE-3.5-4K-0205": 0.012 * RMB,
+ "ERNIE-Speed-8K": 0.004 * RMB,
+ "ERNIE-Speed-128K": 0.004 * RMB,
+ "ERNIE-Lite-8K-0922": 0.008 * RMB,
+ "ERNIE-Lite-8K-0308": 0.003 * RMB,
+ "ERNIE-Tiny-8K": 0.001 * RMB,
+ "BLOOMZ-7B": 0.004 * RMB,
+ "Embedding-V1": 0.002 * RMB,
+ "bge-large-zh": 0.002 * RMB,
+ "bge-large-en": 0.002 * RMB,
+ "tao-8k": 0.002 * RMB,
+ "PaLM-2": 1,
+ "gemini-1.5-pro-latest": 1.25, // $3.5 / 1M tokens
+ "gemini-1.5-flash-latest": 0.075,
+ "gemini-2.0-flash": 0.05,
+ "gemini-2.5-pro-exp-03-25": 0.625,
+ "gemini-2.5-pro-preview-03-25": 0.625,
+ "gemini-2.5-flash-preview-04-17": 0.075,
+ "gemini-2.5-flash-preview-04-17-thinking": 0.075,
+ "gemini-2.5-flash-preview-04-17-nothinking": 0.075,
+ "text-embedding-004": 0.001,
+ "chatglm_turbo": 0.3572, // ￥0.005 / 1k tokens
+ "chatglm_pro": 0.7143, // ￥0.01 / 1k tokens
+ "chatglm_std": 0.3572, // ￥0.005 / 1k tokens
+ "chatglm_lite": 0.1429, // ￥0.002 / 1k tokens
+ "glm-4": 7.143, // ￥0.1 / 1k tokens
+ "glm-4v": 0.05 * RMB, // ￥0.05 / 1k tokens
+ "glm-4-alltools": 0.1 * RMB, // ￥0.1 / 1k tokens
+ "glm-3-turbo": 0.3572,
+ "glm-4-plus": 0.05 * RMB,
+ "glm-4-0520": 0.1 * RMB,
+ "glm-4-air": 0.001 * RMB,
+ "glm-4-airx": 0.01 * RMB,
+ "glm-4-long": 0.001 * RMB,
+ "glm-4-flash": 0,
+ "glm-4v-plus": 0.01 * RMB,
+ "qwen-turbo": 0.8572, // ￥0.012 / 1k tokens
+ "qwen-plus": 10, // ￥0.14 / 1k tokens
+ "text-embedding-v1": 0.05, // ￥0.0007 / 1k tokens
+ "SparkDesk-v1.1": 1.2858, // ￥0.018 / 1k tokens
+ "SparkDesk-v2.1": 1.2858, // ￥0.018 / 1k tokens
+ "SparkDesk-v3.1": 1.2858, // ￥0.018 / 1k tokens
+ "SparkDesk-v3.5": 1.2858, // ￥0.018 / 1k tokens
+ "SparkDesk-v4.0": 1.2858,
+ "360GPT_S2_V9": 0.8572, // ¥0.012 / 1k tokens
+ "360gpt-turbo": 0.0858, // ¥0.0012 / 1k tokens
+ "360gpt-turbo-responsibility-8k": 0.8572, // ¥0.012 / 1k tokens
+ "360gpt-pro": 0.8572, // ¥0.012 / 1k tokens
+ "360gpt2-pro": 0.8572, // ¥0.012 / 1k tokens
+ "embedding-bert-512-v1": 0.0715, // ¥0.001 / 1k tokens
+ "embedding_s1_v1": 0.0715, // ¥0.001 / 1k tokens
+ "semantic_similarity_s1_v1": 0.0715, // ¥0.001 / 1k tokens
+ "hunyuan": 7.143, // ¥0.1 / 1k tokens // https://cloud.tencent.com/document/product/1729/97731#e0e6be58-60c8-469f-bdeb-6c264ce3b4d0
// https://platform.lingyiwanwu.com/docs#-计费单元
// 已经按照 7.2 来换算美元价格
"yi-34b-chat-0205": 0.18,
@@ -275,8 +278,6 @@ func InitModelSettings() {
cacheRatioMapMutex.Lock()
cacheRatioMap = defaultCacheRatio
cacheRatioMapMutex.Unlock()
-
- common.SysLog("model settings initialized")
}
func GetModelPriceMap() map[string]float64 {
@@ -459,6 +460,12 @@ func getHardcodedCompletionModelRatio(name string) (float64, bool) {
return 4, true
} else if strings.HasPrefix(name, "gemini-2.5-pro-preview") {
return 8, true
+ } else if strings.HasPrefix(name, "gemini-2.5-flash-preview") {
+ if strings.HasSuffix(name, "-nothinking") {
+ return 4, false
+ } else {
+ return 3.5 / 0.6, false
+ }
}
return 4, false
}
diff --git a/web/src/components/ModelSetting.js b/web/src/components/ModelSetting.js
index dab001ff..2a566d6b 100644
--- a/web/src/components/ModelSetting.js
+++ b/web/src/components/ModelSetting.js
@@ -20,6 +20,8 @@ const ModelSetting = () => {
'global.pass_through_request_enabled': false,
'general_setting.ping_interval_enabled': false,
'general_setting.ping_interval_seconds': 60,
+ 'gemini.thinking_adapter_enabled': false,
+ 'gemini.thinking_adapter_budget_tokens_percentage': 0.6,
});
let [loading, setLoading] = useState(false);
diff --git a/web/src/pages/Setting/Model/SettingGeminiModel.js b/web/src/pages/Setting/Model/SettingGeminiModel.js
index 6e911080..a8f278db 100644
--- a/web/src/pages/Setting/Model/SettingGeminiModel.js
+++ b/web/src/pages/Setting/Model/SettingGeminiModel.js
@@ -9,6 +9,7 @@ import {
verifyJSON,
} from '../../../helpers';
import { useTranslation } from 'react-i18next';
+import Text from '@douyinfe/semi-ui/lib/es/typography/text.js';
const GEMINI_SETTING_EXAMPLE = {
default: 'OFF',
@@ -27,6 +28,8 @@ export default function SettingGeminiModel(props) {
'gemini.safety_settings': '',
'gemini.version_settings': '',
'gemini.supported_imagine_models': [],
+ 'gemini.thinking_adapter_enabled': false,
+ 'gemini.thinking_adapter_budget_tokens_percentage': 0.6,
});
const refForm = useRef();
const [inputsRow, setInputsRow] = useState(inputs);
@@ -151,13 +154,69 @@ export default function SettingGeminiModel(props) {
/>
+
+
-
+
+
+ {t(
+ "和Claude不同，默认情况下Gemini的思考模型会自动决定要不要思考，就算不开启适配模型也可以正常使用，" +
+ "-nothinking后缀（BudgetTokens=0，思考关闭）也会返回少量的思考token，这是gemini的特性，" +
+ "如果您需要计费，推荐设置无后缀模型价格按思考价格设置"
+ )}
+
+
+
+
+
+
+ setInputs({
+ ...inputs,
+ 'gemini.thinking_adapter_enabled': value,
+ })
+ }
+ />
+
+
+
+
+
+ {t(
+ 'Gemini思考适配 BudgetTokens = MaxTokens * BudgetTokens 百分比',
+ )}
+
+
+
+
+
+
+ setInputs({
+ ...inputs,
+ 'gemini.thinking_adapter_budget_tokens_percentage': value,
+ })
+ }
+ />
+
+
+
+
+