feat: 完善audio计费

common/model-ratio.go

@@ -54,6 +54,7 @@ var defaultModelRatio = map[string]float64{
 	"gpt-4-turbo":                     5, // $0.01 / 1K tokens
 	"gpt-4-turbo-2024-04-09":          5, // $0.01 / 1K tokens
 	//"gpt-3.5-turbo-0301":           0.75, //deprecated
+	"gpt-3.5-turbo":          0.25,
 	"gpt-3.5-turbo-0613":     0.75,
 	"gpt-3.5-turbo-16k":      1.5, // $0.003 / 1K tokens
 	"gpt-3.5-turbo-16k-0613": 1.5,

dto/openai_response.go

@@ -128,7 +128,9 @@ type CompletionsStreamResponse struct {
 }
 
 type Usage struct {
-	PromptTokens     int `json:"prompt_tokens"`
-	CompletionTokens int `json:"completion_tokens"`
-	TotalTokens      int `json:"total_tokens"`
+	PromptTokens           int                `json:"prompt_tokens"`
+	CompletionTokens       int                `json:"completion_tokens"`
+	TotalTokens            int                `json:"total_tokens"`
+	PromptTokensDetails    InputTokenDetails  `json:"prompt_tokens_details"`
+	CompletionTokenDetails OutputTokenDetails `json:"completion_tokens_details"`
 }

dto/realtime.go

@@ -47,6 +47,7 @@ type InputTokenDetails struct {
 	CachedTokens int `json:"cached_tokens"`
 	TextTokens   int `json:"text_tokens"`
 	AudioTokens  int `json:"audio_tokens"`
+	ImageTokens  int `json:"image_tokens"`
 }
 
 type OutputTokenDetails struct {

relay/common/relay_info.go

@@ -41,6 +41,7 @@ type RelayInfo struct {
 	OutputAudioFormat    string
 	RealtimeTools        []dto.RealTimeTool
 	IsFirstRequest       bool
+	AudioUsage           bool
 }
 
 func GenRelayInfoWs(c *gin.Context, ws *websocket.Conn) *RelayInfo {

relay/relay-text.go

@@ -207,7 +207,12 @@ func TextHelper(c *gin.Context) (openaiErr *dto.OpenAIErrorWithStatusCode) {
 		service.ResetStatusCode(openaiErr, statusCodeMappingStr)
 		return openaiErr
 	}
-	postConsumeQuota(c, relayInfo, textRequest.Model, usage.(*dto.Usage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
+
+	if strings.HasPrefix(relayInfo.UpstreamModelName, "gpt-4o-audio") {
+		service.PostAudioConsumeQuota(c, relayInfo, usage.(*dto.Usage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
+	} else {
+		postConsumeQuota(c, relayInfo, textRequest.Model, usage.(*dto.Usage), ratio, preConsumedQuota, userQuota, modelRatio, groupRatio, modelPrice, getModelPriceSuccess, "")
+	}
 	return nil
 }
 

service/log.go

@@ -19,12 +19,26 @@ func GenerateTextOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, m
 	return other
 }
 
-func GenerateWssOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.RealtimeUsage, modelRatio, groupRatio, completionRatio, modelPrice float64) map[string]interface{} {
+func GenerateWssOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.RealtimeUsage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice float64) map[string]interface{} {
 	info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, modelPrice)
 	info["ws"] = true
 	info["audio_input"] = usage.InputTokenDetails.AudioTokens
 	info["audio_output"] = usage.OutputTokenDetails.AudioTokens
 	info["text_input"] = usage.InputTokenDetails.TextTokens
 	info["text_output"] = usage.OutputTokenDetails.TextTokens
+	info["audio_ratio"] = audioRatio
+	info["audio_completion_ratio"] = audioCompletionRatio
+	return info
+}
+
+func GenerateAudioOtherInfo(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage *dto.Usage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice float64) map[string]interface{} {
+	info := GenerateTextOtherInfo(ctx, relayInfo, modelRatio, groupRatio, completionRatio, modelPrice)
+	info["audio"] = true
+	info["audio_input"] = usage.PromptTokensDetails.AudioTokens
+	info["audio_output"] = usage.CompletionTokenDetails.AudioTokens
+	info["text_input"] = usage.PromptTokensDetails.TextTokens
+	info["text_output"] = usage.CompletionTokenDetails.TextTokens
+	info["audio_ratio"] = audioRatio
+	info["audio_completion_ratio"] = audioCompletionRatio
 	return info
 }

service/quota.go

@@ -9,7 +9,6 @@ import (
 	"one-api/dto"
 	"one-api/model"
 	relaycommon "one-api/relay/common"
-	"strings"
 	"time"
 )
 
@@ -124,18 +123,79 @@ func PostWssConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, mod
 	}
 
 	logModel := modelName
-	if strings.HasPrefix(logModel, "gpt-4-gizmo") {
-		logModel = "gpt-4-gizmo-*"
-		logContent += fmt.Sprintf("，模型 %s", modelName)
+	if extraContent != "" {
+		logContent += ", " + extraContent
+	}
+	other := GenerateWssOtherInfo(ctx, relayInfo, usage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice)
+	model.RecordConsumeLog(ctx, relayInfo.UserId, relayInfo.ChannelId, usage.InputTokens, usage.OutputTokens, logModel,
+		tokenName, quota, logContent, relayInfo.TokenId, userQuota, int(useTimeSeconds), relayInfo.IsStream, other)
+}
+
+func PostAudioConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo,
+	usage *dto.Usage, ratio float64, preConsumedQuota int, userQuota int, modelRatio float64,
+	groupRatio float64,
+	modelPrice float64, usePrice bool, extraContent string) {
+
+	useTimeSeconds := time.Now().Unix() - relayInfo.StartTime.Unix()
+	textInputTokens := usage.PromptTokensDetails.TextTokens
+	textOutTokens := usage.CompletionTokenDetails.TextTokens
+
+	audioInputTokens := usage.PromptTokensDetails.AudioTokens
+	audioOutTokens := usage.CompletionTokenDetails.AudioTokens
+
+	tokenName := ctx.GetString("token_name")
+	completionRatio := common.GetCompletionRatio(relayInfo.UpstreamModelName)
+	audioRatio := common.GetAudioRatio(relayInfo.UpstreamModelName)
+	audioCompletionRatio := common.GetAudioCompletionRatio(relayInfo.UpstreamModelName)
+
+	quota := 0
+	if !usePrice {
+		quota = int(math.Round(float64(textInputTokens) + float64(textOutTokens)*completionRatio))
+		quota += int(math.Round(float64(audioInputTokens)*audioRatio + float64(audioOutTokens)*audioRatio*audioCompletionRatio))
+		quota = int(math.Round(float64(quota) * ratio))
+		if ratio != 0 && quota <= 0 {
+			quota = 1
+		}
+	} else {
+		quota = int(modelPrice * common.QuotaPerUnit * groupRatio)
+	}
+	totalTokens := usage.TotalTokens
+	var logContent string
+	if !usePrice {
+		logContent = fmt.Sprintf("模型倍率 %.2f，补全倍率 %.2f，音频倍率 %.2f，音频补全倍率 %.2f，分组倍率 %.2f", modelRatio, completionRatio, audioRatio, audioCompletionRatio, groupRatio)
+	} else {
+		logContent = fmt.Sprintf("模型价格 %.2f，分组倍率 %.2f", modelPrice, groupRatio)
 	}
-	if strings.HasPrefix(logModel, "gpt-4o-gizmo") {
-		logModel = "gpt-4o-gizmo-*"
-		logContent += fmt.Sprintf("，模型 %s", modelName)
+
+	// record all the consume log even if quota is 0
+	if totalTokens == 0 {
+		// in this case, must be some error happened
+		// we cannot just return, because we may have to return the pre-consumed quota
+		quota = 0
+		logContent += fmt.Sprintf("（可能是上游超时）")
+		common.LogError(ctx, fmt.Sprintf("total tokens is 0, cannot consume quota, userId %d, channelId %d, "+
+			"tokenId %d, model %s， pre-consumed quota %d", relayInfo.UserId, relayInfo.ChannelId, relayInfo.TokenId, relayInfo.UpstreamModelName, preConsumedQuota))
+	} else {
+		quotaDelta := quota - preConsumedQuota
+		if quotaDelta != 0 {
+			err := model.PostConsumeTokenQuota(relayInfo, userQuota, quotaDelta, preConsumedQuota, true)
+			if err != nil {
+				common.LogError(ctx, "error consuming token remain quota: "+err.Error())
+			}
+		}
+		err := model.CacheUpdateUserQuota(relayInfo.UserId)
+		if err != nil {
+			common.LogError(ctx, "error update user quota cache: "+err.Error())
+		}
+		model.UpdateUserUsedQuotaAndRequestCount(relayInfo.UserId, quota)
+		model.UpdateChannelUsedQuota(relayInfo.ChannelId, quota)
 	}
+
+	logModel := relayInfo.UpstreamModelName
 	if extraContent != "" {
 		logContent += ", " + extraContent
 	}
-	other := GenerateWssOtherInfo(ctx, relayInfo, usage, modelRatio, groupRatio, completionRatio, modelPrice)
-	model.RecordConsumeLog(ctx, relayInfo.UserId, relayInfo.ChannelId, usage.InputTokens, usage.OutputTokens, logModel,
+	other := GenerateAudioOtherInfo(ctx, relayInfo, usage, modelRatio, groupRatio, completionRatio, audioRatio, audioCompletionRatio, modelPrice)
+	model.RecordConsumeLog(ctx, relayInfo.UserId, relayInfo.ChannelId, usage.PromptTokens, usage.CompletionTokens, logModel,
 		tokenName, quota, logContent, relayInfo.TokenId, userQuota, int(useTimeSeconds), relayInfo.IsStream, other)
 }

web/src/components/LogsTable.js

@@ -24,10 +24,11 @@ import {
 } from '@douyinfe/semi-ui';
 import { ITEMS_PER_PAGE } from '../constants';
 import {
+  renderAudioModelPrice,
   renderModelPrice,
   renderNumber,
   renderQuota,
-  stringToColor,
+  stringToColor
 } from '../helpers/render';
 import Paragraph from '@douyinfe/semi-ui/lib/es/typography/paragraph';
 import { getLogOther } from '../helpers/other.js';
@@ -537,7 +538,7 @@ const LogsTable = () => {
         //   value: content,
         // })
       }
-      if (other?.ws) {
+      if (other?.ws || other?.audio) {
         expandDataLocal.push({
           key: '语音输入',
           value: other.audio_input,
@@ -560,14 +561,30 @@ const LogsTable = () => {
         value: logs[i].content,
       })
       if (logs[i].type === 2) {
-        let content = renderModelPrice(
+        let content = '';
+        if (other?.ws || other?.audio) {
+          content = renderAudioModelPrice(
+            other.text_input,
+            other.text_output,
+            other.model_ratio,
+            other.model_price,
+            other.completion_ratio,
+            other.audio_input,
+            other.audio_output,
+            other?.audio_ratio,
+            other?.audio_completion_ratio,
+            other.group_ratio,
+          );
+        } else {
+          content = renderModelPrice(
             logs[i].prompt_tokens,
             logs[i].completion_tokens,
             other.model_ratio,
             other.model_price,
             other.completion_ratio,
             other.group_ratio,
-        );
+          );
+        }
         expandDataLocal.push({
           key: '计费过程',
           value: content,

web/src/helpers/render.js

@@ -173,6 +173,59 @@ export function renderModelPrice(
   }
 }
 
+export function renderAudioModelPrice(
+  inputTokens,
+  completionTokens,
+  modelRatio,
+  modelPrice = -1,
+  completionRatio,
+  audioInputTokens,
+  audioCompletionTokens,
+  audioRatio,
+  audioCompletionRatio,
+  groupRatio,
+) {
+  // 1 ratio = $0.002 / 1K tokens
+  if (modelPrice !== -1) {
+    return '模型价格：$' + modelPrice + ' * 分组倍率：' + groupRatio + ' = $' + modelPrice * groupRatio;
+  } else {
+    if (completionRatio === undefined) {
+      completionRatio = 0;
+    }
+    // 这里的 *2 是因为 1倍率=0.002刀，请勿删除
+    let inputRatioPrice = modelRatio * 2.0;
+    let completionRatioPrice = modelRatio * 2.0 * completionRatio;
+    let price =
+      (inputTokens / 1000000) * inputRatioPrice * groupRatio +
+      (completionTokens / 1000000) * completionRatioPrice * groupRatio +
+      (audioInputTokens / 1000000) * inputRatioPrice * audioRatio * groupRatio +
+      (audioCompletionTokens / 1000000) * inputRatioPrice * audioRatio * audioCompletionRatio * groupRatio;
+    return (
+      <>
+        <article>
+          <p>提示：${inputRatioPrice} * {groupRatio} = ${inputRatioPrice * groupRatio} / 1M tokens</p>
+          <p>补全：${completionRatioPrice} * {groupRatio} = ${completionRatioPrice * groupRatio} / 1M tokens</p>
+          <p>音频提示：${inputRatioPrice} * {groupRatio} * {audioRatio} = ${inputRatioPrice * audioRatio * groupRatio} / 1M tokens</p>
+          <p>音频补全：${inputRatioPrice} * {groupRatio} * {audioRatio} * {audioCompletionRatio} = ${inputRatioPrice * audioRatio * audioCompletionRatio * groupRatio} / 1M tokens</p>
+          <p></p>
+          <p>
+            提示 {inputTokens} tokens / 1M tokens * ${inputRatioPrice} + 补全{' '}
+            {completionTokens} tokens / 1M tokens * ${completionRatioPrice} +
+          </p>
+          <p>
+            音频提示 {audioInputTokens} tokens / 1M tokens * ${inputRatioPrice} * {audioRatio} + 音频补全 {audioCompletionTokens} tokens / 1M tokens * ${inputRatioPrice} * {audioRatio} * {audioCompletionRatio}
+          </p>
+          <p>
+            （文字 + 音频） * 分组 {groupRatio} =
+            ${price.toFixed(6)}
+          </p>
+          <p>仅供参考，以实际扣费为准</p>
+        </article>
+      </>
+    );
+  }
+}
+
 export function renderQuotaWithPrompt(quota, digits) {
   let displayInCurrency = localStorage.getItem('display_in_currency');
   displayInCurrency = displayInCurrency === 'true';