feat: support baichuan's models now (close #1057)

.github/workflows/linux-release.yml

@@ -23,7 +23,7 @@ jobs:
       - uses: actions/setup-node@v3
         with:
           node-version: 16
-      - name: Build Frontend (theme default)
+      - name: Build Frontend
         env:
           CI: ""
         run: |
@@ -38,7 +38,7 @@ jobs:
       - name: Build Backend (amd64)
         run: |
           go mod download
-          go build -ldflags "-s -w -X 'one-api/common.Version=$(git describe --tags)' -extldflags '-static'" -o one-api
+          go build -ldflags "-s -w -X 'github.com/songquanpeng/one-api/common.Version=$(git describe --tags)' -extldflags '-static'" -o one-api
 
       - name: Build Backend (arm64)
         run: |

.github/workflows/macos-release.yml

@@ -23,7 +23,7 @@ jobs:
       - uses: actions/setup-node@v3
         with:
           node-version: 16
-      - name: Build Frontend (theme default)
+      - name: Build Frontend
         env:
           CI: ""
         run: |
@@ -38,7 +38,7 @@ jobs:
       - name: Build Backend
         run: |
           go mod download
-          go build -ldflags "-X 'one-api/common.Version=$(git describe --tags)'" -o one-api-macos
+          go build -ldflags "-X 'github.com/songquanpeng/one-api/common.Version=$(git describe --tags)'" -o one-api-macos
       - name: Release
         uses: softprops/action-gh-release@v1
         if: startsWith(github.ref, 'refs/tags/')

.github/workflows/windows-release.yml

@@ -26,7 +26,7 @@ jobs:
       - uses: actions/setup-node@v3
         with:
           node-version: 16
-      - name: Build Frontend (theme default)
+      - name: Build Frontend
         env:
           CI: ""
         run: |
@@ -41,7 +41,7 @@ jobs:
       - name: Build Backend
         run: |
           go mod download
-          go build -ldflags "-s -w -X 'one-api/common.Version=$(git describe --tags)'" -o one-api.exe
+          go build -ldflags "-s -w -X 'github.com/songquanpeng/one-api/common.Version=$(git describe --tags)'" -o one-api.exe
       - name: Release
         uses: softprops/action-gh-release@v1
         if: startsWith(github.ref, 'refs/tags/')

Dockerfile

@@ -23,7 +23,7 @@ ADD go.mod go.sum ./
 RUN go mod download
 COPY . .
 COPY --from=builder /web/build ./web/build
-RUN go build -ldflags "-s -w -X 'one-api/common.Version=$(cat VERSION)' -extldflags '-static'" -o one-api
+RUN go build -ldflags "-s -w -X 'github.com/songquanpeng/one-api/common.Version=$(cat VERSION)' -extldflags '-static'" -o one-api
 
 FROM alpine
 

README.md

@@ -74,6 +74,7 @@ _✨ 通过标准的 OpenAI API 格式访问所有的大模型，开箱即用
    + [x] [360 智脑](https://ai.360.cn)
    + [x] [腾讯混元大模型](https://cloud.tencent.com/document/product/1729)
    + [x] [Moonshot AI](https://platform.moonshot.cn/)
+   + [x] [百川大模型](https://platform.baichuan-ai.com)
    + [ ] [字节云雀大模型](https://www.volcengine.com/product/ark) (WIP)
    + [ ] [MINIMAX](https://api.minimax.chat/) (WIP)
 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)。

common/constants.go

@@ -64,6 +64,7 @@ const (
 	ChannelTypeTencent        = 23
 	ChannelTypeGemini         = 24
 	ChannelTypeMoonshot       = 25
+	ChannelTypeBaichuan       = 26
 )
 
 var ChannelBaseURLs = []string{
@@ -93,6 +94,7 @@ var ChannelBaseURLs = []string{
 	"https://hunyuan.cloud.tencent.com",         // 23
 	"https://generativelanguage.googleapis.com", // 24
 	"https://api.moonshot.cn",                   // 25
+	"https://api.baichuan-ai.com",               // 26
 }
 
 const (

common/gin.go

@@ -8,12 +8,24 @@ import (
 	"strings"
 )
 
-func UnmarshalBodyReusable(c *gin.Context, v any) error {
+const KeyRequestBody = "key_request_body"
+
+func GetRequestBody(c *gin.Context) ([]byte, error) {
+	requestBody, _ := c.Get(KeyRequestBody)
+	if requestBody != nil {
+		return requestBody.([]byte), nil
+	}
 	requestBody, err := io.ReadAll(c.Request.Body)
 	if err != nil {
-		return err
+		return nil, err
 	}
-	err = c.Request.Body.Close()
+	_ = c.Request.Body.Close()
+	c.Set(KeyRequestBody, requestBody)
+	return requestBody.([]byte), nil
+}
+
+func UnmarshalBodyReusable(c *gin.Context, v any) error {
+	requestBody, err := GetRequestBody(c)
 	if err != nil {
 		return err
 	}

common/helper/helper.go

@@ -137,6 +137,7 @@ func GetUUID() string {
 }
 
 const keyChars = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"
+const keyNumbers = "0123456789"
 
 func init() {
 	rand.Seed(time.Now().UnixNano())
@@ -168,6 +169,15 @@ func GetRandomString(length int) string {
 	return string(key)
 }
 
+func GetRandomNumberString(length int) string {
+	rand.Seed(time.Now().UnixNano())
+	key := make([]byte, length)
+	for i := 0; i < length; i++ {
+		key[i] = keyNumbers[rand.Intn(len(keyNumbers))]
+	}
+	return string(key)
+}
+
 func GetTimestamp() int64 {
 	return time.Now().Unix()
 }

common/logger/logger.go

@@ -13,6 +13,7 @@ import (
 )
 
 const (
+	loggerDEBUG = "DEBUG"
 	loggerINFO  = "INFO"
 	loggerWarn  = "WARN"
 	loggerError = "ERR"
@@ -55,6 +56,10 @@ func SysError(s string) {
 	_, _ = fmt.Fprintf(gin.DefaultErrorWriter, "[SYS] %v | %s \n", t.Format("2006/01/02 - 15:04:05"), s)
 }
 
+func Debug(ctx context.Context, msg string) {
+	logHelper(ctx, loggerDEBUG, msg)
+}
+
 func Info(ctx context.Context, msg string) {
 	logHelper(ctx, loggerINFO, msg)
 }
@@ -67,6 +72,10 @@ func Error(ctx context.Context, msg string) {
 	logHelper(ctx, loggerError, msg)
 }
 
+func Debugf(ctx context.Context, format string, a ...any) {
+	Debug(ctx, fmt.Sprintf(format, a...))
+}
+
 func Infof(ctx context.Context, format string, a ...any) {
 	Info(ctx, fmt.Sprintf(format, a...))
 }

common/model-ratio.go

@@ -112,6 +112,10 @@ var ModelRatio = map[string]float64{
 	"qwen-max-longcontext":      1.4286, // ￥0.02 / 1k tokens
 	"text-embedding-v1":         0.05,   // ￥0.0007 / 1k tokens
 	"SparkDesk":                 1.2858, // ￥0.018 / 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
 	"360GPT_S2_V9":              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

controller/model.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"github.com/gin-gonic/gin"
 	"github.com/songquanpeng/one-api/relay/channel/ai360"
+	"github.com/songquanpeng/one-api/relay/channel/baichuan"
 	"github.com/songquanpeng/one-api/relay/channel/moonshot"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/helper"
@@ -58,6 +59,9 @@ func init() {
 	})
 	// https://platform.openai.com/docs/models/model-endpoint-compatibility
 	for i := 0; i < constant.APITypeDummy; i++ {
+		if i == constant.APITypeAIProxyLibrary {
+			continue
+		}
 		adaptor := helper.GetAdaptor(i)
 		channelName := adaptor.GetChannelName()
 		modelNames := adaptor.GetModelList()
@@ -95,6 +99,17 @@ func init() {
 			Parent:     nil,
 		})
 	}
+	for _, modelName := range baichuan.ModelList {
+		openAIModels = append(openAIModels, OpenAIModels{
+			Id:         modelName,
+			Object:     "model",
+			Created:    1626777600,
+			OwnedBy:    "baichuan",
+			Permission: permission,
+			Root:       modelName,
+			Parent:     nil,
+		})
+	}
 	openAIModelsMap = make(map[string]OpenAIModels)
 	for _, model := range openAIModels {
 		openAIModelsMap[model.Id] = model

controller/relay.go

@@ -1,23 +1,27 @@
 package controller
 
 import (
+	"bytes"
+	"context"
 	"fmt"
 	"github.com/gin-gonic/gin"
+	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/common/config"
 	"github.com/songquanpeng/one-api/common/helper"
 	"github.com/songquanpeng/one-api/common/logger"
+	"github.com/songquanpeng/one-api/middleware"
+	dbmodel "github.com/songquanpeng/one-api/model"
 	"github.com/songquanpeng/one-api/relay/constant"
 	"github.com/songquanpeng/one-api/relay/controller"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/util"
+	"io"
 	"net/http"
-	"strconv"
 )
 
 // https://platform.openai.com/docs/api-reference/chat
 
-func Relay(c *gin.Context) {
-	relayMode := constant.Path2RelayMode(c.Request.URL.Path)
+func relay(c *gin.Context, relayMode int) *model.ErrorWithStatusCode {
 	var err *model.ErrorWithStatusCode
 	switch relayMode {
 	case constant.RelayModeImagesGenerations:
@@ -31,32 +35,89 @@ func Relay(c *gin.Context) {
 	default:
 		err = controller.RelayTextHelper(c)
 	}
-	if err != nil {
-		requestId := c.GetString(logger.RequestIdKey)
-		retryTimesStr := c.Query("retry")
-		retryTimes, _ := strconv.Atoi(retryTimesStr)
-		if retryTimesStr == "" {
-			retryTimes = config.RetryTimes
+	return err
+}
+
+func Relay(c *gin.Context) {
+	ctx := c.Request.Context()
+	relayMode := constant.Path2RelayMode(c.Request.URL.Path)
+	if config.DebugEnabled {
+		requestBody, _ := common.GetRequestBody(c)
+		logger.Debugf(ctx, "request body: %s", string(requestBody))
+	}
+	bizErr := relay(c, relayMode)
+	if bizErr == nil {
+		return
+	}
+	channelId := c.GetInt("channel_id")
+	lastFailedChannelId := channelId
+	channelName := c.GetString("channel_name")
+	group := c.GetString("group")
+	originalModel := c.GetString("original_model")
+	go processChannelRelayError(ctx, channelId, channelName, bizErr)
+	requestId := c.GetString(logger.RequestIdKey)
+	retryTimes := config.RetryTimes
+	if !shouldRetry(c, bizErr.StatusCode) {
+		logger.Errorf(ctx, "relay error happen, status code is %d, won't retry in this case", bizErr.StatusCode)
+		retryTimes = 0
+	}
+	for i := retryTimes; i > 0; i-- {
+		channel, err := dbmodel.CacheGetRandomSatisfiedChannel(group, originalModel)
+		if err != nil {
+			logger.Errorf(ctx, "CacheGetRandomSatisfiedChannel failed: %w", err)
+			break
 		}
-		if retryTimes > 0 {
-			c.Redirect(http.StatusTemporaryRedirect, fmt.Sprintf("%s?retry=%d", c.Request.URL.Path, retryTimes-1))
-		} else {
-			if err.StatusCode == http.StatusTooManyRequests {
-				err.Error.Message = "当前分组上游负载已饱和，请稍后再试"
-			}
-			err.Error.Message = helper.MessageWithRequestId(err.Error.Message, requestId)
-			c.JSON(err.StatusCode, gin.H{
-				"error": err.Error,
-			})
+		logger.Infof(ctx, "using channel #%d to retry (remain times %d)", channel.Id, i)
+		if channel.Id == lastFailedChannelId {
+			continue
+		}
+		middleware.SetupContextForSelectedChannel(c, channel, originalModel)
+		requestBody, err := common.GetRequestBody(c)
+		c.Request.Body = io.NopCloser(bytes.NewBuffer(requestBody))
+		bizErr = relay(c, relayMode)
+		if bizErr == nil {
+			return
 		}
 		channelId := c.GetInt("channel_id")
-		logger.Error(c.Request.Context(), fmt.Sprintf("relay error (channel #%d): %s", channelId, err.Message))
-		// https://platform.openai.com/docs/guides/error-codes/api-errors
-		if util.ShouldDisableChannel(&err.Error, err.StatusCode) {
-			channelId := c.GetInt("channel_id")
-			channelName := c.GetString("channel_name")
-			disableChannel(channelId, channelName, err.Message)
+		lastFailedChannelId = channelId
+		channelName := c.GetString("channel_name")
+		go processChannelRelayError(ctx, channelId, channelName, bizErr)
+	}
+	if bizErr != nil {
+		if bizErr.StatusCode == http.StatusTooManyRequests {
+			bizErr.Error.Message = "当前分组上游负载已饱和，请稍后再试"
 		}
+		bizErr.Error.Message = helper.MessageWithRequestId(bizErr.Error.Message, requestId)
+		c.JSON(bizErr.StatusCode, gin.H{
+			"error": bizErr.Error,
+		})
+	}
+}
+
+func shouldRetry(c *gin.Context, statusCode int) bool {
+	if _, ok := c.Get("specific_channel_id"); ok {
+		return false
+	}
+	if statusCode == http.StatusTooManyRequests {
+		return true
+	}
+	if statusCode/100 == 5 {
+		return true
+	}
+	if statusCode == http.StatusBadRequest {
+		return false
+	}
+	if statusCode/100 == 2 {
+		return false
+	}
+	return true
+}
+
+func processChannelRelayError(ctx context.Context, channelId int, channelName string, err *model.ErrorWithStatusCode) {
+	logger.Errorf(ctx, "relay error (channel #%d): %s", channelId, err.Message)
+	// https://platform.openai.com/docs/guides/error-codes/api-errors
+	if util.ShouldDisableChannel(&err.Error, err.StatusCode) {
+		disableChannel(channelId, channelName, err.Message)
 	}
 }
 

i18n/en.json

@@ -456,6 +456,7 @@
   "已绑定的邮箱账户": "Email Account Bound",
   "用户信息更新成功！": "User information updated successfully!",
   "模型倍率 %.2f，分组倍率 %.2f": "model rate %.2f, group rate %.2f",
+  "模型倍率 %.2f，分组倍率 %.2f，补全倍率 %.2f": "model rate %.2f, group rate %.2f, completion rate %.2f",
   "使用明细（总消耗额度：{renderQuota(stat.quota)}）": "Usage Details (Total Consumption Quota: {renderQuota(stat.quota)})",
   "用户名称": "User Name",
   "令牌名称": "Token Name",

middleware/auth.go

@@ -108,7 +108,7 @@ func TokenAuth() func(c *gin.Context) {
 		c.Set("token_name", token.Name)
 		if len(parts) > 1 {
 			if model.IsAdmin(token.UserId) {
-				c.Set("channelId", parts[1])
+				c.Set("specific_channel_id", parts[1])
 			} else {
 				abortWithMessage(c, http.StatusForbidden, "普通用户不支持指定渠道")
 				return

middleware/distributor.go

@@ -21,8 +21,9 @@ func Distribute() func(c *gin.Context) {
 		userId := c.GetInt("id")
 		userGroup, _ := model.CacheGetUserGroup(userId)
 		c.Set("group", userGroup)
+		var requestModel string
 		var channel *model.Channel
-		channelId, ok := c.Get("channelId")
+		channelId, ok := c.Get("specific_channel_id")
 		if ok {
 			id, err := strconv.Atoi(channelId.(string))
 			if err != nil {
@@ -66,6 +67,7 @@ func Distribute() func(c *gin.Context) {
 					modelRequest.Model = "whisper-1"
 				}
 			}
+			requestModel = modelRequest.Model
 			channel, err = model.CacheGetRandomSatisfiedChannel(userGroup, modelRequest.Model)
 			if err != nil {
 				message := fmt.Sprintf("当前分组 %s 下对于模型 %s 无可用渠道", userGroup, modelRequest.Model)
@@ -77,29 +79,34 @@ func Distribute() func(c *gin.Context) {
 				return
 			}
 		}
-		c.Set("channel", channel.Type)
-		c.Set("channel_id", channel.Id)
-		c.Set("channel_name", channel.Name)
-		c.Set("model_mapping", channel.GetModelMapping())
-		c.Request.Header.Set("Authorization", fmt.Sprintf("Bearer %s", channel.Key))
-		c.Set("base_url", channel.GetBaseURL())
-		// this is for backward compatibility
-		switch channel.Type {
-		case common.ChannelTypeAzure:
-			c.Set(common.ConfigKeyAPIVersion, channel.Other)
-		case common.ChannelTypeXunfei:
-			c.Set(common.ConfigKeyAPIVersion, channel.Other)
-		case common.ChannelTypeGemini:
-			c.Set(common.ConfigKeyAPIVersion, channel.Other)
-		case common.ChannelTypeAIProxyLibrary:
-			c.Set(common.ConfigKeyLibraryID, channel.Other)
-		case common.ChannelTypeAli:
-			c.Set(common.ConfigKeyPlugin, channel.Other)
-		}
-		cfg, _ := channel.LoadConfig()
-		for k, v := range cfg {
-			c.Set(common.ConfigKeyPrefix+k, v)
-		}
+		SetupContextForSelectedChannel(c, channel, requestModel)
 		c.Next()
 	}
 }
+
+func SetupContextForSelectedChannel(c *gin.Context, channel *model.Channel, modelName string) {
+	c.Set("channel", channel.Type)
+	c.Set("channel_id", channel.Id)
+	c.Set("channel_name", channel.Name)
+	c.Set("model_mapping", channel.GetModelMapping())
+	c.Set("original_model", modelName) // for retry
+	c.Request.Header.Set("Authorization", fmt.Sprintf("Bearer %s", channel.Key))
+	c.Set("base_url", channel.GetBaseURL())
+	// this is for backward compatibility
+	switch channel.Type {
+	case common.ChannelTypeAzure:
+		c.Set(common.ConfigKeyAPIVersion, channel.Other)
+	case common.ChannelTypeXunfei:
+		c.Set(common.ConfigKeyAPIVersion, channel.Other)
+	case common.ChannelTypeGemini:
+		c.Set(common.ConfigKeyAPIVersion, channel.Other)
+	case common.ChannelTypeAIProxyLibrary:
+		c.Set(common.ConfigKeyLibraryID, channel.Other)
+	case common.ChannelTypeAli:
+		c.Set(common.ConfigKeyPlugin, channel.Other)
+	}
+	cfg, _ := channel.LoadConfig()
+	for k, v := range cfg {
+		c.Set(common.ConfigKeyPrefix+k, v)
+	}
+}

middleware/request-id.go

@@ -9,7 +9,7 @@ import (
 
 func RequestId() func(c *gin.Context) {
 	return func(c *gin.Context) {
-		id := helper.GetTimeString() + helper.GetRandomString(8)
+		id := helper.GetTimeString() + helper.GetRandomNumberString(8)
 		c.Set(logger.RequestIdKey, id)
 		ctx := context.WithValue(c.Request.Context(), logger.RequestIdKey, id)
 		c.Request = c.Request.WithContext(ctx)

model/cache.go

@@ -94,7 +94,7 @@ func CacheUpdateUserQuota(id int) error {
 	if !common.RedisEnabled {
 		return nil
 	}
-	quota, err := GetUserQuota(id)
+	quota, err := CacheGetUserQuota(id)
 	if err != nil {
 		return err
 	}

relay/channel/aiproxy/main.go

@@ -53,7 +53,7 @@ func responseAIProxyLibrary2OpenAI(response *LibraryResponse) *openai.TextRespon
 		FinishReason: "stop",
 	}
 	fullTextResponse := openai.TextResponse{
-		Id:      helper.GetUUID(),
+		Id:      fmt.Sprintf("chatcmpl-%s", helper.GetUUID()),
 		Object:  "chat.completion",
 		Created: helper.GetTimestamp(),
 		Choices: []openai.TextResponseChoice{choice},
@@ -66,7 +66,7 @@ func documentsAIProxyLibrary(documents []LibraryDocument) *openai.ChatCompletion
 	choice.Delta.Content = aiProxyDocuments2Markdown(documents)
 	choice.FinishReason = &constant.StopFinishReason
 	return &openai.ChatCompletionsStreamResponse{
-		Id:      helper.GetUUID(),
+		Id:      fmt.Sprintf("chatcmpl-%s", helper.GetUUID()),
 		Object:  "chat.completion.chunk",
 		Created: helper.GetTimestamp(),
 		Model:   "",
@@ -78,7 +78,7 @@ func streamResponseAIProxyLibrary2OpenAI(response *LibraryStreamResponse) *opena
 	var choice openai.ChatCompletionsStreamResponseChoice
 	choice.Delta.Content = response.Content
 	return &openai.ChatCompletionsStreamResponse{
-		Id:      helper.GetUUID(),
+		Id:      fmt.Sprintf("chatcmpl-%s", helper.GetUUID()),
 		Object:  "chat.completion.chunk",
 		Created: helper.GetTimestamp(),
 		Model:   response.Model,

relay/channel/baichuan/constants.go

@@ -0,0 +1,7 @@
+package baichuan
+
+var ModelList = []string{
+	"Baichuan2-Turbo",
+	"Baichuan2-Turbo-192k",
+	"Baichuan-Text-Embedding",
+}

relay/channel/openai/adaptor.go

@@ -7,6 +7,7 @@ import (
 	"github.com/songquanpeng/one-api/common"
 	"github.com/songquanpeng/one-api/relay/channel"
 	"github.com/songquanpeng/one-api/relay/channel/ai360"
+	"github.com/songquanpeng/one-api/relay/channel/baichuan"
 	"github.com/songquanpeng/one-api/relay/channel/moonshot"
 	"github.com/songquanpeng/one-api/relay/model"
 	"github.com/songquanpeng/one-api/relay/util"
@@ -84,6 +85,8 @@ func (a *Adaptor) GetModelList() []string {
 		return ai360.ModelList
 	case common.ChannelTypeMoonshot:
 		return moonshot.ModelList
+	case common.ChannelTypeBaichuan:
+		return baichuan.ModelList
 	default:
 		return ModelList
 	}
@@ -97,6 +100,8 @@ func (a *Adaptor) GetChannelName() string {
 		return "360"
 	case common.ChannelTypeMoonshot:
 		return "moonshot"
+	case common.ChannelTypeBaichuan:
+		return "baichuan"
 	default:
 		return "openai"
 	}

relay/channel/openai/model.go

@@ -118,8 +118,10 @@ type ImageResponse struct {
 }
 
 type ChatCompletionsStreamResponseChoice struct {
+	Index int `json:"index"`
 	Delta struct {
 		Content string `json:"content"`
+		Role    string `json:"role,omitempty"`
 	} `json:"delta"`
 	FinishReason *string `json:"finish_reason,omitempty"`
 }

relay/channel/xunfei/constants.go

@@ -2,4 +2,8 @@ package xunfei
 
 var ModelList = []string{
 	"SparkDesk",
+	"SparkDesk-v1.1",
+	"SparkDesk-v2.1",
+	"SparkDesk-v3.1",
+	"SparkDesk-v3.5",
 }

relay/channel/xunfei/main.go

@@ -70,6 +70,7 @@ func responseXunfei2OpenAI(response *ChatResponse) *openai.TextResponse {
 		FinishReason: constant.StopFinishReason,
 	}
 	fullTextResponse := openai.TextResponse{
+		Id:      fmt.Sprintf("chatcmpl-%s", helper.GetUUID()),
 		Object:  "chat.completion",
 		Created: helper.GetTimestamp(),
 		Choices: []openai.TextResponseChoice{choice},
@@ -92,6 +93,7 @@ func streamResponseXunfei2OpenAI(xunfeiResponse *ChatResponse) *openai.ChatCompl
 		choice.FinishReason = &constant.StopFinishReason
 	}
 	response := openai.ChatCompletionsStreamResponse{
+		Id:      fmt.Sprintf("chatcmpl-%s", helper.GetUUID()),
 		Object:  "chat.completion.chunk",
 		Created: helper.GetTimestamp(),
 		Model:   "SparkDesk",
@@ -127,7 +129,7 @@ func buildXunfeiAuthUrl(hostUrl string, apiKey, apiSecret string) string {
 }
 
 func StreamHandler(c *gin.Context, textRequest model.GeneralOpenAIRequest, appId string, apiSecret string, apiKey string) (*model.ErrorWithStatusCode, *model.Usage) {
-	domain, authUrl := getXunfeiAuthUrl(c, apiKey, apiSecret)
+	domain, authUrl := getXunfeiAuthUrl(c, apiKey, apiSecret, textRequest.Model)
 	dataChan, stopChan, err := xunfeiMakeRequest(textRequest, domain, authUrl, appId)
 	if err != nil {
 		return openai.ErrorWrapper(err, "make xunfei request err", http.StatusInternalServerError), nil
@@ -157,7 +159,7 @@ func StreamHandler(c *gin.Context, textRequest model.GeneralOpenAIRequest, appId
 }
 
 func Handler(c *gin.Context, textRequest model.GeneralOpenAIRequest, appId string, apiSecret string, apiKey string) (*model.ErrorWithStatusCode, *model.Usage) {
-	domain, authUrl := getXunfeiAuthUrl(c, apiKey, apiSecret)
+	domain, authUrl := getXunfeiAuthUrl(c, apiKey, apiSecret, textRequest.Model)
 	dataChan, stopChan, err := xunfeiMakeRequest(textRequest, domain, authUrl, appId)
 	if err != nil {
 		return openai.ErrorWrapper(err, "make xunfei request err", http.StatusInternalServerError), nil
@@ -242,20 +244,45 @@ func xunfeiMakeRequest(textRequest model.GeneralOpenAIRequest, domain, authUrl,
 	return dataChan, stopChan, nil
 }
 
-func getXunfeiAuthUrl(c *gin.Context, apiKey string, apiSecret string) (string, string) {
+func getAPIVersion(c *gin.Context, modelName string) string {
 	query := c.Request.URL.Query()
 	apiVersion := query.Get("api-version")
-	if apiVersion == "" {
-		apiVersion = c.GetString(common.ConfigKeyAPIVersion)
+	if apiVersion != "" {
+		return apiVersion
 	}
-	if apiVersion == "" {
-		apiVersion = "v1.1"
-		logger.SysLog("api_version not found, use default: " + apiVersion)
+	parts := strings.Split(modelName, "-")
+	if len(parts) == 2 {
+		apiVersion = parts[1]
+		return apiVersion
+
 	}
-	domain := "general"
-	if apiVersion != "v1.1" {
-		domain += strings.Split(apiVersion, ".")[0]
+	apiVersion = c.GetString(common.ConfigKeyAPIVersion)
+	if apiVersion != "" {
+		return apiVersion
 	}
+	apiVersion = "v1.1"
+	logger.SysLog("api_version not found, using default: " + apiVersion)
+	return apiVersion
+}
+
+// https://www.xfyun.cn/doc/spark/Web.html#_1-%E6%8E%A5%E5%8F%A3%E8%AF%B4%E6%98%8E
+func apiVersion2domain(apiVersion string) string {
+	switch apiVersion {
+	case "v1.1":
+		return "general"
+	case "v2.1":
+		return "generalv2"
+	case "v3.1":
+		return "generalv3"
+	case "v3.5":
+		return "generalv3.5"
+	}
+	return "general" + apiVersion
+}
+
+func getXunfeiAuthUrl(c *gin.Context, apiKey string, apiSecret string, modelName string) (string, string) {
+	apiVersion := getAPIVersion(c, modelName)
+	domain := apiVersion2domain(apiVersion)
 	authUrl := buildXunfeiAuthUrl(fmt.Sprintf("wss://spark-api.xf-yun.com/%s/chat", apiVersion), apiKey, apiSecret)
 	return domain, authUrl
 }

relay/controller/text.go

@@ -39,6 +39,7 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
 	ratio := modelRatio * groupRatio
 	// pre-consume quota
 	promptTokens := getPromptTokens(textRequest, meta.Mode)
+	meta.PromptTokens = promptTokens
 	preConsumedQuota, bizErr := preConsumeQuota(ctx, textRequest, promptTokens, ratio, meta)
 	if bizErr != nil {
 		logger.Warnf(ctx, "preConsumeQuota failed: %+v", *bizErr)
@@ -54,7 +55,8 @@ func RelayTextHelper(c *gin.Context) *model.ErrorWithStatusCode {
 	var requestBody io.Reader
 	if meta.APIType == constant.APITypeOpenAI {
 		// no need to convert request for openai
-		if isModelMapped {
+		shouldResetRequestBody := isModelMapped || meta.ChannelType == common.ChannelTypeBaichuan // frequency_penalty 0 is not acceptable for baichuan
+		if shouldResetRequestBody {
 			jsonStr, err := json.Marshal(textRequest)
 			if err != nil {
 				return openai.ErrorWrapper(err, "json_marshal_failed", http.StatusInternalServerError)

web/THEMES

@@ -1,2 +1,2 @@
 default
-berry
+berry

web/berry/src/constants/ChannelConstants.js

@@ -71,6 +71,12 @@ export const CHANNEL_OPTIONS = {
     value: 23,
     color: 'default'
   },
+  26: {
+    key: 26,
+    text: '百川大模型',
+    value: 23,
+    color: 'default'
+  },
   8: {
     key: 8,
     text: '自定义渠道',

web/berry/src/views/Channel/index.js

@@ -227,9 +227,9 @@ export default function ChannelPage() {
                 <Button onClick={testAllChannels} startIcon={<IconBrandSpeedtest width={'18px'} />}>
                   测试启用渠道
                 </Button>
-                <Button onClick={updateAllChannelsBalance} startIcon={<IconCoinYuan width={'18px'} />}>
-                  更新启用余额
-                </Button>
+                {/*<Button onClick={updateAllChannelsBalance} startIcon={<IconCoinYuan width={'18px'} />}>*/}
+                {/*  更新启用余额*/}
+                {/*</Button>*/}
                 <Button onClick={deleteAllDisabledChannels} startIcon={<IconHttpDelete width={'18px'} />}>
                   删除禁用渠道
                 </Button>

web/berry/src/views/Channel/type/Config.js

@@ -94,7 +94,13 @@ const typeConfig = {
       other: "版本号",
     },
     input: {
-      models: ["SparkDesk"],
+      models: [
+          "SparkDesk",
+        'SparkDesk-v1.1',
+        'SparkDesk-v2.1',
+        'SparkDesk-v3.1',
+        'SparkDesk-v3.5'
+      ],
     },
     prompt: {
       key: "按照如下格式输入：APPID|APISecret|APIKey",
@@ -139,6 +145,18 @@ const typeConfig = {
     },
     modelGroup: "google gemini",
   },
+  25: {
+    input: {
+      models: ['moonshot-v1-8k', 'moonshot-v1-32k', 'moonshot-v1-128k'],
+    },
+    modelGroup: "moonshot",
+  },
+  26: {
+    input: {
+      models: ['Baichuan2-Turbo', 'Baichuan2-Turbo-192k', 'Baichuan-Text-Embedding'],
+    },
+    modelGroup: "baichuan",
+  },
 };
 
 export { defaultConfig, typeConfig };

web/build.sh

@@ -1,13 +1,13 @@
 #!/bin/sh
 
 version=$(cat VERSION)
-themes=$(cat THEMES)
-IFS=$'\n'
+pwd
 
-for theme in $themes; do
+while IFS= read -r theme; do
     echo "Building theme: $theme"
-    cd $theme
+    rm -r build/$theme
+    cd "$theme"
     npm install
     DISABLE_ESLINT_PLUGIN='true' REACT_APP_VERSION=$version npm run build
     cd ..
-done
+done < THEMES

web/default/src/components/ChannelsTable.js

@@ -522,8 +522,8 @@ const ChannelsTable = () => {
               <Button size='small' loading={loading} onClick={testAllChannels}>
                 测试所有渠道
               </Button>
-              <Button size='small' onClick={updateAllChannelsBalance}
-                      loading={loading || updatingBalance}>更新已启用渠道余额</Button>
+              {/*<Button size='small' onClick={updateAllChannelsBalance}*/}
+              {/*        loading={loading || updatingBalance}>更新已启用渠道余额</Button>*/}
               <Popup
                 trigger={
                   <Button size='small' loading={loading}>

web/default/src/constants/channel.constants.js

@@ -11,6 +11,7 @@ export const CHANNEL_OPTIONS = [
   { key: 19, text: '360 智脑', value: 19, color: 'blue' },
   { key: 25, text: 'Moonshot AI', value: 25, color: 'black' },
   { key: 23, text: '腾讯混元', value: 23, color: 'teal' },
+  { key: 26, text: '百川大模型', value: 26, color: 'orange' },
   { key: 8, text: '自定义渠道', value: 8, color: 'pink' },
   { key: 22, text: '知识库：FastGPT', value: 22, color: 'blue' },
   { key: 21, text: '知识库：AI Proxy', value: 21, color: 'purple' },

web/default/src/pages/Channel/EditChannel.js

@@ -82,7 +82,13 @@ const EditChannel = () => {
           localModels = ['chatglm_turbo', 'chatglm_pro', 'chatglm_std', 'chatglm_lite'];
           break;
         case 18:
-          localModels = ['SparkDesk'];
+          localModels = [
+            'SparkDesk',
+            'SparkDesk-v1.1',
+            'SparkDesk-v2.1',
+            'SparkDesk-v3.1',
+            'SparkDesk-v3.5'
+          ];
           break;
         case 19:
           localModels = ['360GPT_S2_V9', 'embedding-bert-512-v1', 'embedding_s1_v1', 'semantic_similarity_s1_v1'];
@@ -96,6 +102,9 @@ const EditChannel = () => {
         case 25:
           localModels = ['moonshot-v1-8k', 'moonshot-v1-32k', 'moonshot-v1-128k'];
           break;
+        case 26:
+          localModels = ['Baichuan2-Turbo', 'Baichuan2-Turbo-192k', 'Baichuan-Text-Embedding'];
+          break;
       }
       setInputs((inputs) => ({ ...inputs, models: localModels }));
     }