Add Default Options Value Override Feature

README.md

@@ -90,7 +90,6 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用 
    + [x] [together.ai](https://www.together.ai/)
    + [x] [novita.ai](https://www.novita.ai/)
    + [x] [硅基流动 SiliconCloud](https://siliconflow.cn/siliconcloud)
-   + [x] [xAI](https://x.ai/)
 2. 支持配置镜像以及众多[第三方代理服务](https://iamazing.cn/page/openai-api-third-party-services)。
 3. 支持通过**负载均衡**的方式访问多个渠道。
 4. 支持 **stream 模式**，可以通过流式传输实现打字机效果。

common/ctxkey/key.go

@@ -14,6 +14,7 @@ const (
 	OriginalModel     = "original_model"
 	Group             = "group"
 	ModelMapping      = "model_mapping"
+	ParamsOverride    = "params_override"
 	ChannelName       = "channel_name"
 	TokenId           = "token_id"
 	TokenName         = "token_name"

controller/channel-test.go

@@ -76,9 +76,9 @@
 		if len(modelNames) > 0 {
 			modelName = modelNames[0]
 		}
-	}
-	if modelMap != nil && modelMap[modelName] != "" {
-		modelName = modelMap[modelName]
+		if modelMap != nil && modelMap[modelName] != "" {
+			modelName = modelMap[modelName]
+		}
 	}
 	meta.OriginModelName, meta.ActualModelName = request.Model, modelName
 	request.Model = modelName

middleware/distributor.go

@@ -62,6 +62,7 @@
 	c.Set(ctxkey.ChannelId, channel.Id)
 	c.Set(ctxkey.ChannelName, channel.Name)
 	c.Set(ctxkey.ModelMapping, channel.GetModelMapping())
+	c.Set(ctxkey.ParamsOverride, channel.GetParamsOverride())
 	c.Set(ctxkey.OriginalModel, modelName) // for retry
 	c.Request.Header.Set("Authorization", fmt.Sprintf("Bearer %s", channel.Key))
 	c.Set(ctxkey.BaseURL, channel.GetBaseURL())

model/channel.go

@@ -35,6 +35,7 @@
 	Group              string  `json:"group" gorm:"type:varchar(32);default:'default'"`
 	UsedQuota          int64   `json:"used_quota" gorm:"bigint;default:0"`
 	ModelMapping       *string `json:"model_mapping" gorm:"type:varchar(1024);default:''"`
+	ParamsOverride     *string `json:"default_params_override" gorm:"type:text;default:''"`
 	Priority           *int64  `json:"priority" gorm:"bigint;default:0"`
 	Config             string  `json:"config"`
 }
@@ -123,6 +124,20 @@
 	return modelMapping
 }
 
+func (channel *Channel) GetParamsOverride() map[string]map[string]interface{} {
+    if channel.ParamsOverride == nil || *channel.ParamsOverride == "" || *channel.ParamsOverride == "{}" {
+        return nil
+    }
+    paramsOverride := make(map[string]map[string]interface{})
+    err := json.Unmarshal([]byte(*channel.ParamsOverride), &paramsOverride)
+    if err != nil {
+        logger.SysError(fmt.Sprintf("failed to unmarshal params override for channel %d, error: %s", channel.Id, err.Error()))
+        return nil
+    }
+    return paramsOverride
+}
+
+
 func (channel *Channel) Insert() error {
 	var err error
 	err = DB.Create(channel).Error

relay/adaptor/anthropic/constants.go

@@ -3,9 +3,7 @@ package anthropic
 var ModelList = []string{
 	"claude-instant-1.2", "claude-2.0", "claude-2.1",
 	"claude-3-haiku-20240307",
-	"claude-3-5-haiku-20241022",
 	"claude-3-sonnet-20240229",
 	"claude-3-opus-20240229",
 	"claude-3-5-sonnet-20240620",
-	"claude-3-5-sonnet-20241022",
 }

relay/adaptor/aws/claude/main.go

@@ -31,10 +31,8 @@ var AwsModelIDMap = map[string]string{
 	"claude-2.1":                 "anthropic.claude-v2:1",
 	"claude-3-sonnet-20240229":   "anthropic.claude-3-sonnet-20240229-v1:0",
 	"claude-3-5-sonnet-20240620": "anthropic.claude-3-5-sonnet-20240620-v1:0",
-	"claude-3-5-sonnet-20241022": "anthropic.claude-3-5-sonnet-20241022-v2:0",
 	"claude-3-opus-20240229":     "anthropic.claude-3-opus-20240229-v1:0",
 	"claude-3-haiku-20240307":    "anthropic.claude-3-haiku-20240307-v1:0",
-	"claude-3-5-haiku-20241022":  "anthropic.claude-3-5-haiku-20241022-v1:0",
 }
 
 func awsModelID(requestModel string) (string, error) {

relay/adaptor/gemini/main.go

@@ -4,12 +4,11 @@ import (
 	"bufio"
 	"encoding/json"
 	"fmt"
+	"github.com/songquanpeng/one-api/common/render"
 	"io"
 	"net/http"
 	"strings"
 
-	"github.com/songquanpeng/one-api/common/render"
-
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
@@ -29,11 +28,6 @@ const (
 	VisionMaxImageNum = 16
 )
 
-var mimeTypeMap = map[string]string{
-	"json_object": "application/json",
-	"text":        "text/plain",
-}
-
 // Setting safety to the lowest possible values since Gemini is already powerless enough
 func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
 	geminiRequest := ChatRequest{
@@ -62,15 +56,6 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
 			MaxOutputTokens: textRequest.MaxTokens,
 		},
 	}
-	if textRequest.ResponseFormat != nil {
-		if mimeType, ok := mimeTypeMap[textRequest.ResponseFormat.Type]; ok {
-			geminiRequest.GenerationConfig.ResponseMimeType = mimeType
-		}
-		if textRequest.ResponseFormat.JsonSchema != nil {
-			geminiRequest.GenerationConfig.ResponseSchema = textRequest.ResponseFormat.JsonSchema.Schema
-			geminiRequest.GenerationConfig.ResponseMimeType = mimeTypeMap["json_object"]
-		}
-	}
 	if textRequest.Tools != nil {
 		functions := make([]model.Function, 0, len(textRequest.Tools))
 		for _, tool := range textRequest.Tools {

relay/adaptor/gemini/model.go

@@ -65,12 +65,10 @@ type ChatTools struct {
 }
 
 type ChatGenerationConfig struct {
-	ResponseMimeType string   `json:"responseMimeType,omitempty"`
-	ResponseSchema   any      `json:"responseSchema,omitempty"`
-	Temperature      float64  `json:"temperature,omitempty"`
-	TopP             float64  `json:"topP,omitempty"`
-	TopK             float64  `json:"topK,omitempty"`
-	MaxOutputTokens  int      `json:"maxOutputTokens,omitempty"`
-	CandidateCount   int      `json:"candidateCount,omitempty"`
-	StopSequences    []string `json:"stopSequences,omitempty"`
+	Temperature     float64  `json:"temperature,omitempty"`
+	TopP            float64  `json:"topP,omitempty"`
+	TopK            float64  `json:"topK,omitempty"`
+	MaxOutputTokens int      `json:"maxOutputTokens,omitempty"`
+	CandidateCount  int      `json:"candidateCount,omitempty"`
+	StopSequences   []string `json:"stopSequences,omitempty"`
 }

relay/adaptor/groq/constants.go

@@ -4,21 +4,14 @@ package groq
 
 var ModelList = []string{
 	"gemma-7b-it",
+	"mixtral-8x7b-32768",
+	"llama3-8b-8192",
+	"llama3-70b-8192",
 	"gemma2-9b-it",
+	"llama-3.1-405b-reasoning",
 	"llama-3.1-70b-versatile",
 	"llama-3.1-8b-instant",
-	"llama-3.2-11b-text-preview",
-	"llama-3.2-11b-vision-preview",
-	"llama-3.2-1b-preview",
-	"llama-3.2-3b-preview",
-	"llama-3.2-90b-text-preview",
-	"llama-guard-3-8b",
-	"llama3-70b-8192",
-	"llama3-8b-8192",
 	"llama3-groq-70b-8192-tool-use-preview",
 	"llama3-groq-8b-8192-tool-use-preview",
-	"llava-v1.5-7b-4096-preview",
-	"mixtral-8x7b-32768",
-	"distil-whisper-large-v3-en",
 	"whisper-large-v3",
 }

relay/adaptor/openai/adaptor.go

@@ -75,13 +75,6 @@ func (a *Adaptor) ConvertRequest(c *gin.Context, relayMode int, request *model.G
 	if request == nil {
 		return nil, errors.New("request is nil")
 	}
-	if request.Stream {
-		// always return usage in stream mode
-		if request.StreamOptions == nil {
-			request.StreamOptions = &model.StreamOptions{}
-		}
-		request.StreamOptions.IncludeUsage = true
-	}
 	return request, nil
 }
 

relay/adaptor/openai/compatible.go

@@ -11,10 +11,9 @@ import (
 	"github.com/songquanpeng/one-api/relay/adaptor/mistral"
 	"github.com/songquanpeng/one-api/relay/adaptor/moonshot"
 	"github.com/songquanpeng/one-api/relay/adaptor/novita"
-	"github.com/songquanpeng/one-api/relay/adaptor/siliconflow"
 	"github.com/songquanpeng/one-api/relay/adaptor/stepfun"
 	"github.com/songquanpeng/one-api/relay/adaptor/togetherai"
-	"github.com/songquanpeng/one-api/relay/adaptor/xai"
+	"github.com/songquanpeng/one-api/relay/adaptor/siliconflow"
 	"github.com/songquanpeng/one-api/relay/channeltype"
 )
 
@@ -33,7 +32,6 @@ var CompatibleChannels = []int{
 	channeltype.TogetherAI,
 	channeltype.Novita,
 	channeltype.SiliconFlow,
-	channeltype.XAI,
 }
 
 func GetCompatibleChannelMeta(channelType int) (string, []string) {
@@ -66,8 +64,6 @@ func GetCompatibleChannelMeta(channelType int) (string, []string) {
 		return "novita", novita.ModelList
 	case channeltype.SiliconFlow:
 		return "siliconflow", siliconflow.ModelList
-	case channeltype.XAI:
-		return "xai", xai.ModelList
 	default:
 		return "openai", ModelList
 	}

relay/adaptor/vertexai/gemini/adapter.go

@@ -15,7 +15,7 @@ import (
 )
 
 var ModelList = []string{
-	"gemini-1.5-pro-001", "gemini-1.5-flash-001", "gemini-pro", "gemini-pro-vision", "gemini-1.5-pro-002", "gemini-1.5-flash-002", 
+	"gemini-1.5-pro-001", "gemini-1.5-flash-001", "gemini-pro", "gemini-pro-vision",
 }
 
 type Adaptor struct {

relay/adaptor/xunfei/constants.go

@@ -7,6 +7,5 @@ var ModelList = []string{
 	"SparkDesk-v3.1",
 	"SparkDesk-v3.1-128K",
 	"SparkDesk-v3.5",
-	"SparkDesk-v3.5-32K",
 	"SparkDesk-v4.0",
 }

relay/adaptor/xunfei/main.go

@@ -292,8 +292,6 @@
 		return "pro-128k"
 	case "v3.5":
 		return "generalv3.5"
-	case "v3.5-32K":
-		return "max-32k"
 	case "v4.0":
 		return "4.0Ultra"
 	}
@@ -305,10 +303,7 @@
 	domain := apiVersion2domain(apiVersion)
 	switch apiVersion {
 	case "v3.1-128K":
-		authUrl = buildXunfeiAuthUrl(fmt.Sprintf("wss://spark-api.xf-yun.com/chat/pro-128k"), apiKey, apiSecret)
-		break
-	case "v3.5-32K":
-		authUrl = buildXunfeiAuthUrl(fmt.Sprintf("wss://spark-api.xf-yun.com/chat/max-32k"), apiKey, apiSecret)
+		authUrl = buildXunfeiAuthUrl(fmt.Sprintf("wss://spark-api.xf-yun.com/%s/pro-128k", apiVersion), apiKey, apiSecret)
 		break
 	default:
 		authUrl = buildXunfeiAuthUrl(fmt.Sprintf("wss://spark-api.xf-yun.com/%s/chat", apiVersion), apiKey, apiSecret)

relay/billing/ratio/model.go

@@ -79,10 +79,8 @@ var ModelRatio = map[string]float64{
 	"claude-2.0":                 8.0 / 1000 * USD,
 	"claude-2.1":                 8.0 / 1000 * USD,
 	"claude-3-haiku-20240307":    0.25 / 1000 * USD,
-	"claude-3-5-haiku-20241022":  1.0 / 1000 * USD,
 	"claude-3-sonnet-20240229":   3.0 / 1000 * USD,
 	"claude-3-5-sonnet-20240620": 3.0 / 1000 * USD,
-	"claude-3-5-sonnet-20241022": 3.0 / 1000 * USD,
 	"claude-3-opus-20240229":     15.0 / 1000 * USD,
 	// https://cloud.baidu.com/doc/WENXINWORKSHOP/s/hlrk4akp7
 	"ERNIE-4.0-8K":       0.120 * RMB,
@@ -132,7 +130,6 @@ var ModelRatio = map[string]float64{
 	"SparkDesk-v3.1":            1.2858, // ￥0.018 / 1k tokens
 	"SparkDesk-v3.1-128K":       1.2858, // ￥0.018 / 1k tokens
 	"SparkDesk-v3.5":            1.2858, // ￥0.018 / 1k tokens
-	"SparkDesk-v3.5-32K":        1.2858, // ￥0.018 / 1k tokens
 	"SparkDesk-v4.0":            1.2858, // ￥0.018 / 1k tokens
 	"360GPT_S2_V9":              0.8572, // ¥0.012 / 1k tokens
 	"embedding-bert-512-v1":     0.0715, // ¥0.001 / 1k tokens
@@ -164,21 +161,15 @@ var ModelRatio = map[string]float64{
 	"mistral-embed":         0.1 / 1000 * USD,
 	// https://wow.groq.com/#:~:text=inquiries%C2%A0here.-,Model,-Current%20Speed
 	"gemma-7b-it":                           0.07 / 1000000 * USD,
+	"mixtral-8x7b-32768":                    0.24 / 1000000 * USD,
+	"llama3-8b-8192":                        0.05 / 1000000 * USD,
+	"llama3-70b-8192":                       0.59 / 1000000 * USD,
 	"gemma2-9b-it":                          0.20 / 1000000 * USD,
+	"llama-3.1-405b-reasoning":              0.89 / 1000000 * USD,
 	"llama-3.1-70b-versatile":               0.59 / 1000000 * USD,
 	"llama-3.1-8b-instant":                  0.05 / 1000000 * USD,
-	"llama-3.2-11b-text-preview":            0.05 / 1000000 * USD,
-	"llama-3.2-11b-vision-preview":          0.05 / 1000000 * USD,
-	"llama-3.2-1b-preview":                  0.05 / 1000000 * USD,
-	"llama-3.2-3b-preview":                  0.05 / 1000000 * USD,
-	"llama-3.2-90b-text-preview":            0.59 / 1000000 * USD,
-	"llama-guard-3-8b":                      0.05 / 1000000 * USD,
-	"llama3-70b-8192":                       0.59 / 1000000 * USD,
-	"llama3-8b-8192":                        0.05 / 1000000 * USD,
 	"llama3-groq-70b-8192-tool-use-preview": 0.89 / 1000000 * USD,
 	"llama3-groq-8b-8192-tool-use-preview":  0.19 / 1000000 * USD,
-	"mixtral-8x7b-32768":                    0.24 / 1000000 * USD,
-
 	// https://platform.lingyiwanwu.com/docs#-计费单元
 	"yi-34b-chat-0205": 2.5 / 1000 * RMB,
 	"yi-34b-chat-200k": 12.0 / 1000 * RMB,
@@ -209,8 +200,6 @@ var ModelRatio = map[string]float64{
 	"deepl-zh": 25.0 / 1000 * USD,
 	"deepl-en": 25.0 / 1000 * USD,
 	"deepl-ja": 25.0 / 1000 * USD,
-	// https://console.x.ai/
-	"grok-beta": 5.0 / 1000 * USD,
 }
 
 var CompletionRatio = map[string]float64{
@@ -375,8 +364,6 @@ func GetCompletionRatio(name string, channelType int) float64 {
 		return 3
 	case "command-r-plus":
 		return 5
-	case "grok-beta":
-		return 3
 	}
 	return 1
 }

relay/channeltype/define.go

@@ -46,6 +46,5 @@ const (
 	VertextAI
 	Proxy
 	SiliconFlow
-	XAI
 	Dummy
 )

relay/channeltype/url.go

@@ -45,8 +45,7 @@ var ChannelBaseURLs = []string{
 	"https://api.novita.ai/v3/openai",           // 41
 	"",                                          // 42
 	"",                                          // 43
-	"https://api.siliconflow.cn",                // 44
-	"https://api.x.ai",                          // 45
+	"https://api.siliconflow.cn",                 // 44
 }
 
 func init() {