-
-## 🏘️ 社区交流群
-
-扫码加入飞书话题群：
-
-
-
-
-
-
-
## 🏘️ 加入我们
我们正在寻找志同道合的小伙伴，加速 FastGPT 的发展。你可以通过 [FastGPT 2025 招聘](https://fael3z0zfze.feishu.cn/wiki/P7FOwEmPziVcaYkvVaacnVX1nvg)了解 FastGPT 的招聘信息。
@@ -135,17 +125,25 @@ https://github.com/labring/FastGPT/assets/15308462/7d3a38df-eb0e-4388-9250-2409b
- [Sealos：快速部署集群应用](https://github.com/labring/sealos)
- [AI Proxy API调用地址](https://sealos.run/aiproxy/?k=fastgpt-github/)
- [One API：多模型管理，支持 Azure、文心一言等](https://github.com/songquanpeng/one-api)
-- [TuShan：5 分钟搭建后台管理系统](https://github.com/msgbyte/tushan)
-
## 🌿 第三方生态
-- [COW 个人微信/企微机器人](https://doc.tryfastgpt.ai/docs/use-cases/external-integration/onwechat/)
- [SiliconCloud (硅基流动) —— 开源模型在线体验平台](https://cloud.siliconflow.cn/i/TR9Ym0c4)
+- [COW 个人微信/企微机器人](https://doc.tryfastgpt.ai/docs/use-cases/external-integration/onwechat/)
+
+
+
+
+
+## 🏘️ 社区交流群
+
+扫码加入飞书话题群：
+
+
diff --git a/deploy/docker/docker-compose-milvus.yml b/deploy/docker/docker-compose-milvus.yml
index 2914ab87a..2e30c9f23 100644
--- a/deploy/docker/docker-compose-milvus.yml
+++ b/deploy/docker/docker-compose-milvus.yml
@@ -141,10 +141,9 @@ services:
- AIPROXY_API_ENDPOINT=http://aiproxy:3000
# AI Proxy 的 Admin Token，与 AI Proxy 中的环境变量 ADMIN_KEY
- AIPROXY_API_TOKEN=aiproxy
- # AI模型的API地址哦。务必加 /v1。这里默认填写了OneApi的访问地址。
- - OPENAI_BASE_URL=http://oneapi:3000/v1
- # AI模型的API Key。（这里默认填写了OneAPI的快速默认key，测试通后，务必及时修改）
- - CHAT_API_KEY=sk-fastgpt
+ # 模型中转地址（如果用了 AI Proxy，下面 2 个就不需要了，旧版 OneAPI 用户，使用下面的变量）
+ # - OPENAI_BASE_URL=http://oneapi:3000/v1
+ # - CHAT_API_KEY=sk-fastgpt
# 数据库最大连接数
- DB_MAX_LINK=30
# 登录凭证密钥
@@ -180,32 +179,37 @@ services:
container_name: aiproxy
restart: unless-stopped
depends_on:
- pgsql:
+ aiproxy_pg:
condition: service_healthy
ports:
- - '3002:3000/tcp'
+ - '3002:3000'
networks:
- fastgpt
environment:
- - ADMIN_KEY=aiproxy # 对应 fastgpt 里的AIPROXY_API_TOKEN
- - LOG_DETAIL_STORAGE_HOURS=1 # 日志详情保存时间（小时）
- - TZ=Asia/Shanghai
- - SQL_DSN=postgres://postgres:aiproxy@pgsql:5432/aiproxy
+ # 对应 fastgpt 里的AIPROXY_API_TOKEN
+ - ADMIN_KEY=aiproxy
+ # 错误日志详情保存时间（小时）
+ - LOG_DETAIL_STORAGE_HOURS=1
+ # 数据库连接地址
+ - SQL_DSN=postgres://postgres:aiproxy@aiproxy_pg:5432/aiproxy
+ # 最大重试次数
+ - RetryTimes=3
+ # 不需要计费
+ - BILLING_ENABLED=false
+ # 不需要严格检测模型
- DISABLE_MODEL_CONFIG=true
healthcheck:
test: ['CMD', 'curl', '-f', 'http://localhost:3000/api/status']
interval: 5s
timeout: 5s
retries: 10
-
- # AI Proxy
- pgsql:
- # image: "postgres:latest"
- image: registry.cn-hangzhou.aliyuncs.com/fastgpt/pgvector:v0.7.0 # 阿里云
+ aiproxy_pg:
+ # image: pgvector/pgvector:0.8.0-pg15 # docker hub
+ image: registry.cn-hangzhou.aliyuncs.com/fastgpt/pgvector:v0.8.0-pg15 # 阿里云
restart: unless-stopped
- container_name: pgsql
+ container_name: aiproxy_pg
volumes:
- - ./pgsql:/var/lib/postgresql/data
+ - ./aiproxy_pg:/var/lib/postgresql/data
networks:
- fastgpt
environment:
diff --git a/deploy/docker/docker-compose-pgvector.yml b/deploy/docker/docker-compose-pgvector.yml
index c403741bf..aee2bebcb 100644
--- a/deploy/docker/docker-compose-pgvector.yml
+++ b/deploy/docker/docker-compose-pgvector.yml
@@ -11,8 +11,8 @@ services:
# image: registry.cn-hangzhou.aliyuncs.com/fastgpt/pgvector:v0.8.0-pg15 # 阿里云
container_name: pg
restart: always
- ports: # 生产环境建议不要暴露
- - 5432:5432
+ # ports: # 生产环境建议不要暴露
+ # - 5432:5432
networks:
- fastgpt
environment:
@@ -99,10 +99,9 @@ services:
- AIPROXY_API_ENDPOINT=http://aiproxy:3000
# AI Proxy 的 Admin Token，与 AI Proxy 中的环境变量 ADMIN_KEY
- AIPROXY_API_TOKEN=aiproxy
- # AI模型的API地址哦。务必加 /v1。这里默认填写了OneApi的访问地址。
- - OPENAI_BASE_URL=http://oneapi:3000/v1
- # AI模型的API Key。（这里默认填写了OneAPI的快速默认key，测试通后，务必及时修改）
- - CHAT_API_KEY=sk-fastgpt
+ # 模型中转地址（如果用了 AI Proxy，下面 2 个就不需要了，旧版 OneAPI 用户，使用下面的变量）
+ # - OPENAI_BASE_URL=http://oneapi:3000/v1
+ # - CHAT_API_KEY=sk-fastgpt
# 数据库最大连接数
- DB_MAX_LINK=30
# 登录凭证密钥
@@ -137,32 +136,37 @@ services:
container_name: aiproxy
restart: unless-stopped
depends_on:
- pgsql:
+ aiproxy_pg:
condition: service_healthy
ports:
- - '3002:3000/tcp'
+ - '3002:3000'
networks:
- fastgpt
environment:
- - ADMIN_KEY=aiproxy # 对应 fastgpt 里的AIPROXY_API_TOKEN
- - LOG_DETAIL_STORAGE_HOURS=1 # 日志详情保存时间（小时）
- - TZ=Asia/Shanghai
- - SQL_DSN=postgres://postgres:aiproxy@pgsql:5432/aiproxy
+ # 对应 fastgpt 里的AIPROXY_API_TOKEN
+ - ADMIN_KEY=aiproxy
+ # 错误日志详情保存时间（小时）
+ - LOG_DETAIL_STORAGE_HOURS=1
+ # 数据库连接地址
+ - SQL_DSN=postgres://postgres:aiproxy@aiproxy_pg:5432/aiproxy
+ # 最大重试次数
+ - RetryTimes=3
+ # 不需要计费
+ - BILLING_ENABLED=false
+ # 不需要严格检测模型
- DISABLE_MODEL_CONFIG=true
healthcheck:
test: ['CMD', 'curl', '-f', 'http://localhost:3000/api/status']
interval: 5s
timeout: 5s
retries: 10
-
- # AI Proxy
- pgsql:
- # image: "postgres:latest"
- image: registry.cn-hangzhou.aliyuncs.com/fastgpt/pgvector:v0.7.0 # 阿里云
+ aiproxy_pg:
+ # image: pgvector/pgvector:0.8.0-pg15 # docker hub
+ image: registry.cn-hangzhou.aliyuncs.com/fastgpt/pgvector:v0.8.0-pg15 # 阿里云
restart: unless-stopped
- container_name: pgsql
+ container_name: aiproxy_pg
volumes:
- - ./pgsql:/var/lib/postgresql/data
+ - ./aiproxy_pg:/var/lib/postgresql/data
networks:
- fastgpt
environment:
diff --git a/deploy/docker/docker-compose-zilliz.yml b/deploy/docker/docker-compose-zilliz.yml
index ffdb623c1..e2bad44d8 100644
--- a/deploy/docker/docker-compose-zilliz.yml
+++ b/deploy/docker/docker-compose-zilliz.yml
@@ -79,10 +79,9 @@ services:
- AIPROXY_API_ENDPOINT=http://aiproxy:3000
# AI Proxy 的 Admin Token，与 AI Proxy 中的环境变量 ADMIN_KEY
- AIPROXY_API_TOKEN=aiproxy
- # AI模型的API地址哦。务必加 /v1。这里默认填写了OneApi的访问地址。
- - OPENAI_BASE_URL=http://oneapi:3000/v1
- # AI模型的API Key。（这里默认填写了OneAPI的快速默认key，测试通后，务必及时修改）
- - CHAT_API_KEY=sk-fastgpt
+ # 模型中转地址（如果用了 AI Proxy，下面 2 个就不需要了，旧版 OneAPI 用户，使用下面的变量）
+ # - OPENAI_BASE_URL=http://oneapi:3000/v1
+ # - CHAT_API_KEY=sk-fastgpt
# 数据库最大连接数
- DB_MAX_LINK=30
# 登录凭证密钥
@@ -118,32 +117,37 @@ services:
container_name: aiproxy
restart: unless-stopped
depends_on:
- pgsql:
+ aiproxy_pg:
condition: service_healthy
ports:
- - '3002:3000/tcp'
+ - '3002:3000'
networks:
- fastgpt
environment:
- - ADMIN_KEY=aiproxy # 对应 fastgpt 里的AIPROXY_API_TOKEN
- - LOG_DETAIL_STORAGE_HOURS=1 # 日志详情保存时间（小时）
- - TZ=Asia/Shanghai
- - SQL_DSN=postgres://postgres:aiproxy@pgsql:5432/aiproxy
+ # 对应 fastgpt 里的AIPROXY_API_TOKEN
+ - ADMIN_KEY=aiproxy
+ # 错误日志详情保存时间（小时）
+ - LOG_DETAIL_STORAGE_HOURS=1
+ # 数据库连接地址
+ - SQL_DSN=postgres://postgres:aiproxy@aiproxy_pg:5432/aiproxy
+ # 最大重试次数
+ - RetryTimes=3
+ # 不需要计费
+ - BILLING_ENABLED=false
+ # 不需要严格检测模型
- DISABLE_MODEL_CONFIG=true
healthcheck:
test: ['CMD', 'curl', '-f', 'http://localhost:3000/api/status']
interval: 5s
timeout: 5s
retries: 10
-
- # AI Proxy
- pgsql:
- # image: "postgres:latest"
- image: registry.cn-hangzhou.aliyuncs.com/fastgpt/pgvector:v0.7.0 # 阿里云
+ aiproxy_pg:
+ # image: pgvector/pgvector:0.8.0-pg15 # docker hub
+ image: registry.cn-hangzhou.aliyuncs.com/fastgpt/pgvector:v0.8.0-pg15 # 阿里云
restart: unless-stopped
- container_name: pgsql
+ container_name: aiproxy_pg
volumes:
- - ./pgsql:/var/lib/postgresql/data
+ - ./aiproxy_pg:/var/lib/postgresql/data
networks:
- fastgpt
environment:
diff --git a/docSite/assets/imgs/aiproxy1.png b/docSite/assets/imgs/aiproxy1.png
new file mode 100644
index 000000000..3905df329
Binary files /dev/null and b/docSite/assets/imgs/aiproxy1.png differ
diff --git a/docSite/content/zh-cn/docs/development/docker.md b/docSite/content/zh-cn/docs/development/docker.md
index 0ef3b1dcc..9a5962a04 100644
--- a/docSite/content/zh-cn/docs/development/docker.md
+++ b/docSite/content/zh-cn/docs/development/docker.md
@@ -30,7 +30,7 @@ weight: 707
### PgVector版本
-非常轻量，适合数据量在 5000 万以下。
+非常轻量，适合知识库索引量在 5000 万以下。
{{< table "table-hover table-striped-columns" >}}
| 环境 | 最低配置（单节点） | 推荐配置 |
@@ -149,18 +149,14 @@ curl -o docker-compose.yml https://raw.githubusercontent.com/labring/FastGPT/mai
{{< tab tabName="PgVector版本" >}}
{{< markdownify >}}
-```
-FE_DOMAIN=你的前端你访问地址,例如 http://192.168.0.1:3000;https://cloud.fastgpt.cn
-```
+无需操作
{{< /markdownify >}}
{{< /tab >}}
{{< tab tabName="Milvus版本" >}}
{{< markdownify >}}
-```
-FE_DOMAIN=你的前端你访问地址,例如 http://192.168.0.1:3000;https://cloud.fastgpt.cn
-```
+无需操作
{{< /markdownify >}}
{{< /tab >}}
@@ -174,7 +170,6 @@ FE_DOMAIN=你的前端你访问地址,例如 http://192.168.0.1:3000;https://clo
{{% alert icon="🤖" context="success" %}}
1. 修改`MILVUS_ADDRESS`和`MILVUS_TOKEN`链接参数，分别对应 `zilliz` 的 `Public Endpoint` 和 `Api key`，记得把自己ip加入白名单。
-2. 修改FE_DOMAIN=你的前端你访问地址,例如 http://192.168.0.1:3000;https://cloud.fastgpt.cn
{{% /alert %}}
@@ -189,30 +184,28 @@ FE_DOMAIN=你的前端你访问地址,例如 http://192.168.0.1:3000;https://clo
```bash
# 启动容器
docker-compose up -d
-# 等待10s，OneAPI第一次总是要重启几次才能连上Mysql
-sleep 10
-# 重启一次oneapi(由于OneAPI的默认Key有点问题，不重启的话会提示找不到渠道，临时手动重启一次解决，等待作者修复)
-docker restart oneapi
```
### 4. 访问 FastGPT
-目前可以通过 `ip:3000` 直接访问(注意防火墙)。登录用户名为 `root`，密码为`docker-compose.yml`环境变量里设置的 `DEFAULT_ROOT_PSW`。
+目前可以通过 `ip:3000` 直接访问(注意开放防火墙)。登录用户名为 `root`，密码为`docker-compose.yml`环境变量里设置的 `DEFAULT_ROOT_PSW`。
如果需要域名访问，请自行安装并配置 Nginx。
-首次运行，会自动初始化 root 用户，密码为 `1234`（与环境变量中的`DEFAULT_ROOT_PSW`一致），日志里会提示一次`MongoServerError: Unable to read from a snapshot due to pending collection catalog changes;`可忽略。
+首次运行，会自动初始化 root 用户，密码为 `1234`（与环境变量中的`DEFAULT_ROOT_PSW`一致），日志可能会提示一次`MongoServerError: Unable to read from a snapshot due to pending collection catalog changes;`可忽略。
### 5. 配置模型
-登录FastGPT后，进入“模型提供商”页面，首先配置模型渠道，[点击查看相关教程](/docs/development/modelconfig/ai-proxy)
-
-然后配置具体模型，务必先配置至少一个语言模型和一个向量模型，否则系统无法正常使用。
-
-[点击查看模型配置教程](/docs/development/modelConfig/intro/)
+- 首次登录FastGPT后，系统会提示未配置`语言模型`和`索引模型`，并自动跳转模型配置页面。系统必须至少有这两类模型才能正常使用。
+- 如果系统未正常跳转，可以在`账号-模型提供商`页面，进行模型配置。[点击查看相关教程](/docs/development/modelconfig/ai-proxy)
+- 目前已知可能问题：首次进入系统后，整个浏览器 tab 无法响应。此时需要删除该tab，重新打开一次即可。
## FAQ
+### 登录系统后，浏览器无法响应
+
+无法点击任何内容，刷新也无效。此时需要删除该tab，重新打开一次即可。
+
### Mongo 副本集自动初始化失败
最新的 docker-compose 示例优化 Mongo 副本集初始化，实现了全自动。目前在 unbuntu20,22 centos7, wsl2, mac, window 均通过测试。仍无法正常启动，大部分是因为 cpu 不支持 AVX 指令集，可以切换 Mongo4.x 版本。
diff --git a/docSite/content/zh-cn/docs/development/modelConfig/ai-proxy.md b/docSite/content/zh-cn/docs/development/modelConfig/ai-proxy.md
index 70e58881f..5d86c0957 100644
--- a/docSite/content/zh-cn/docs/development/modelConfig/ai-proxy.md
+++ b/docSite/content/zh-cn/docs/development/modelConfig/ai-proxy.md
@@ -7,7 +7,7 @@ toc: true
weight: 744
---
-从 FastGPT 4.8.23 版本开始，引入 AI Proxy 来进一步方便模型的配置。
+从 `FastGPT 4.8.23` 版本开始，引入 AI Proxy 来进一步方便模型的配置。
AI Proxy 与 One API 类似，也是作为一个 OpenAI 接口管理 & 分发系统，可以通过标准的 OpenAI API 格式访问所有的大模型，开箱即用。
@@ -15,13 +15,29 @@ AI Proxy 与 One API 类似，也是作为一个 OpenAI 接口管理 & 分发系
### Docker 版本
-`docker-compose.yml` 文件已加入了 AI Proxy 配置，可直接使用。
+`docker-compose.yml` 文件已加入了 AI Proxy 配置，可直接使用。[点击查看最新的 yml 配置](https://raw.githubusercontent.com/labring/FastGPT/main/deploy/docker/docker-compose-pgvector.yml)
-## 基础使用
+从旧版升级的用户，可以复制 yml 里，ai proxy 的配置，加入到旧的 yml 文件中。
+
+## 运行原理
+
+AI proxy 核心模块:
+
+1. 渠道管理：管理各家模型提供商的 API Key 和可用模型列表。
+2. 模型调用：根据请求的模型，选中对应的渠道；根据渠道的 API 格式，构造请求体，发送请求；格式化响应体成标准格式返回。
+3. 调用日志：详细记录模型调用的日志，并在错误时候可以记录其入参和报错信息，方便排查。
+
+运行流程：
+
+
+
+## 在 FastGPT 中使用
+
+AI proxy 相关功能，可以在`账号-模型提供商`页面找到。
### 1. 创建渠道
-如果 FastGPT 的环境变量中，设置了 AIPROXY_API_ENDPOINT 的值，那么在“模型提供商”的配置页面，会多出两个 tab，可以直接在 FastGPT 平台上配置模型渠道，以及查看模型实际调用日志。
+在`模型提供商`的配置页面，点击`模型渠道`，进入渠道配置页面
@@ -36,9 +52,18 @@ AI Proxy 与 One API 类似，也是作为一个 OpenAI 接口管理 & 分发系
1. 渠道名：展示在外部的渠道名称，仅作标识;
2. 厂商：模型对应的厂商，不同厂商对应不同的默认地址和 API 密钥格式;
3. 模型：当前渠道具体可以使用的模型，系统内置了主流的一些模型，如果下拉框中没有想要的选项，可以点击“新增模型”，[增加自定义模型](/docs/development/modelconfig/intro/#新增自定义模型);
-4. 模型映射：将 FastGPT 请求的模型，映射到具体提供的模型上；
-5. 代理地址：具体请求的地址，系统给每个主流渠道配置了默认的地址，如果无需改动则不用填
-6. API 密钥：从模型厂商处获取的 API 凭证
+4. 模型映射：将 FastGPT 请求的模型，映射到具体提供的模型上。例如：
+
+```json
+{
+ "gpt-4o-test": "gpt-4o",
+}
+```
+
+FatGPT 中的模型为 `gpt-4o-test`，向 AI Proxy 发起请求时也是 `gpt-4o-test`。AI proxy 在向上游发送请求时，实际的`model`为 `gpt-4o`。
+
+5. 代理地址：具体请求的地址，系统给每个主流渠道配置了默认的地址，如果无需改动则不用填。
+6. API 密钥：从模型厂商处获取的 API 凭证。注意部分厂商需要提供多个密钥组合，可以根据提示进行输入。
最后点击“新增”，就能在“模型渠道”下看到刚刚配置的渠道
@@ -60,16 +85,15 @@ AI Proxy 与 One API 类似，也是作为一个 OpenAI 接口管理 & 分发系
### 3. 启用模型
-最后在“模型配置”中，可以选择启用对应的模型，这样就能在平台中使用了
+最后在`模型配置`中，可以选择启用对应的模型，这样就能在平台中使用了，更多模型配置可以参考[模型配置](/docs/development/modelconfig/intro)
-
-## 渠道设置
+## 其他功能介绍
### 优先级
-在 FastGPT 中，可以给渠道设置优先级，对于同样的模型，优先级越高的渠道会越优先请求
+范围1～100。数值越大，越容易被优先选中。
@@ -81,13 +105,15 @@ AI Proxy 与 One API 类似，也是作为一个 OpenAI 接口管理 & 分发系
### 调用日志
-在 “调用日志” 页面，会展示发送到模型处的请求记录，包括具体的输入输出 tokens、请求时间、请求耗时、请求地址等等
+在 `调用日志` 页面，会展示发送到模型处的请求记录，包括具体的输入输出 tokens、请求时间、请求耗时、请求地址等等。错误的请求，则会详细的入参和错误信息，方便排查，但仅会保留 1 小时(环境变量里可配置)。
-## 如何从 OneAPI 迁移到 AI Proxy
+## 从 OneAPI 迁移到 AI Proxy
-可以从任意终端，发起 1 个 HTTP 请求。其中 {{host}} 替换成 AI Proxy 地址，{{admin_key}} 替换成 AI Proxy 中 ADMIN_KEY 的值，参数 dsn 为 OneAPI 的 mysql 连接串
+可以从任意终端，发起 1 个 HTTP 请求。其中 `{{host}}` 替换成 AI Proxy 地址，`{{admin_key}}` 替换成 AI Proxy 中 `ADMIN_KEY` 的值。
+
+Body 参数 `dsn` 为 OneAPI 的 mysql 连接串。
```bash
curl --location --request POST '{{host}}/api/channels/import/oneapi' \
@@ -100,4 +126,4 @@ curl --location --request POST '{{host}}/api/channels/import/oneapi' \
执行成功的情况下会返回 "success": true
-脚本目前不是完全准，可能会有部分渠道遗漏，还需要手动再检查下
\ No newline at end of file
+脚本目前不是完全准，仅是简单的做数据映射，主要是迁移`代理地址`、`模型`和`API 密钥`，建议迁移后再进行手动检查。
\ No newline at end of file
diff --git a/docSite/content/zh-cn/docs/development/upgrading/490.md b/docSite/content/zh-cn/docs/development/upgrading/490.md
index 7836521d5..4857f7068 100644
--- a/docSite/content/zh-cn/docs/development/upgrading/490.md
+++ b/docSite/content/zh-cn/docs/development/upgrading/490.md
@@ -46,6 +46,7 @@ curl --location --request POST 'https://{{host}}/api/admin/initv490' \
1. 知识库数据不再限制索引数量，可无限自定义。同时可自动更新输入文本的索引，不影响自定义索引。
2. Markdown 解析，增加链接后中文标点符号检测，增加空格。
+3. Prompt 模式工具调用，支持思考模型。同时优化其格式检测，减少空输出的概率。
## 🐛 修复
diff --git a/packages/global/core/chat/adapt.ts b/packages/global/core/chat/adapt.ts
index 4c4d65101..4302ca757 100644
--- a/packages/global/core/chat/adapt.ts
+++ b/packages/global/core/chat/adapt.ts
@@ -1,8 +1,11 @@
import type {
+ AIChatItemValueItemType,
ChatItemType,
ChatItemValueItemType,
RuntimeUserPromptType,
- UserChatItemType
+ SystemChatItemValueItemType,
+ UserChatItemType,
+ UserChatItemValueItemType
} from '../../core/chat/type.d';
import { ChatFileTypeEnum, ChatItemValueTypeEnum, ChatRoleEnum } from '../../core/chat/constants';
import type {
@@ -174,137 +177,24 @@ export const GPTMessages2Chats = (
): ChatItemType[] => {
const chatMessages = messages
.map((item) => {
- const value: ChatItemType['value'] = [];
const obj = GPT2Chat[item.role];
- if (
- obj === ChatRoleEnum.System &&
- item.role === ChatCompletionRequestMessageRoleEnum.System
- ) {
- if (Array.isArray(item.content)) {
- item.content.forEach((item) => [
- value.push({
- type: ChatItemValueTypeEnum.text,
- text: {
- content: item.text
- }
- })
- ]);
- } else {
- value.push({
- type: ChatItemValueTypeEnum.text,
- text: {
- content: item.content
- }
- });
- }
- } else if (
- obj === ChatRoleEnum.Human &&
- item.role === ChatCompletionRequestMessageRoleEnum.User
- ) {
- if (typeof item.content === 'string') {
- value.push({
- type: ChatItemValueTypeEnum.text,
- text: {
- content: item.content
- }
- });
- } else if (Array.isArray(item.content)) {
- item.content.forEach((item) => {
- if (item.type === 'text') {
+ const value = (() => {
+ if (
+ obj === ChatRoleEnum.System &&
+ item.role === ChatCompletionRequestMessageRoleEnum.System
+ ) {
+ const value: SystemChatItemValueItemType[] = [];
+
+ if (Array.isArray(item.content)) {
+ item.content.forEach((item) => [
value.push({
type: ChatItemValueTypeEnum.text,
text: {
content: item.text
}
- });
- } else if (item.type === 'image_url') {
- value.push({
- //@ts-ignore
- type: ChatItemValueTypeEnum.file,
- file: {
- type: ChatFileTypeEnum.image,
- name: '',
- url: item.image_url.url
- }
- });
- } else if (item.type === 'file_url') {
- value.push({
- // @ts-ignore
- type: ChatItemValueTypeEnum.file,
- file: {
- type: ChatFileTypeEnum.file,
- name: item.name,
- url: item.url
- }
- });
- }
- });
- }
- } else if (
- obj === ChatRoleEnum.AI &&
- item.role === ChatCompletionRequestMessageRoleEnum.Assistant
- ) {
- if (item.tool_calls && reserveTool) {
- // save tool calls
- const toolCalls = item.tool_calls as ChatCompletionMessageToolCall[];
- value.push({
- //@ts-ignore
- type: ChatItemValueTypeEnum.tool,
- tools: toolCalls.map((tool) => {
- let toolResponse =
- messages.find(
- (msg) =>
- msg.role === ChatCompletionRequestMessageRoleEnum.Tool &&
- msg.tool_call_id === tool.id
- )?.content || '';
- toolResponse =
- typeof toolResponse === 'string' ? toolResponse : JSON.stringify(toolResponse);
-
- return {
- id: tool.id,
- toolName: tool.toolName || '',
- toolAvatar: tool.toolAvatar || '',
- functionName: tool.function.name,
- params: tool.function.arguments,
- response: toolResponse as string
- };
- })
- });
- } else if (item.function_call && reserveTool) {
- const functionCall = item.function_call as ChatCompletionMessageFunctionCall;
- const functionResponse = messages.find(
- (msg) =>
- msg.role === ChatCompletionRequestMessageRoleEnum.Function &&
- msg.name === item.function_call?.name
- ) as ChatCompletionFunctionMessageParam;
-
- if (functionResponse) {
- value.push({
- //@ts-ignore
- type: ChatItemValueTypeEnum.tool,
- tools: [
- {
- id: functionCall.id || '',
- toolName: functionCall.toolName || '',
- toolAvatar: functionCall.toolAvatar || '',
- functionName: functionCall.name,
- params: functionCall.arguments,
- response: functionResponse.content || ''
- }
- ]
- });
- }
- } else if (item.interactive) {
- value.push({
- //@ts-ignore
- type: ChatItemValueTypeEnum.interactive,
- interactive: item.interactive
- });
- } else if (typeof item.content === 'string') {
- const lastValue = value[value.length - 1];
- if (lastValue && lastValue.type === ChatItemValueTypeEnum.text && lastValue.text) {
- lastValue.text.content += item.content;
+ })
+ ]);
} else {
value.push({
type: ChatItemValueTypeEnum.text,
@@ -313,8 +203,145 @@ export const GPTMessages2Chats = (
}
});
}
+ return value;
+ } else if (
+ obj === ChatRoleEnum.Human &&
+ item.role === ChatCompletionRequestMessageRoleEnum.User
+ ) {
+ const value: UserChatItemValueItemType[] = [];
+
+ if (typeof item.content === 'string') {
+ value.push({
+ type: ChatItemValueTypeEnum.text,
+ text: {
+ content: item.content
+ }
+ });
+ } else if (Array.isArray(item.content)) {
+ item.content.forEach((item) => {
+ if (item.type === 'text') {
+ value.push({
+ type: ChatItemValueTypeEnum.text,
+ text: {
+ content: item.text
+ }
+ });
+ } else if (item.type === 'image_url') {
+ value.push({
+ //@ts-ignore
+ type: ChatItemValueTypeEnum.file,
+ file: {
+ type: ChatFileTypeEnum.image,
+ name: '',
+ url: item.image_url.url
+ }
+ });
+ } else if (item.type === 'file_url') {
+ value.push({
+ // @ts-ignore
+ type: ChatItemValueTypeEnum.file,
+ file: {
+ type: ChatFileTypeEnum.file,
+ name: item.name,
+ url: item.url
+ }
+ });
+ }
+ });
+ }
+ return value;
+ } else if (
+ obj === ChatRoleEnum.AI &&
+ item.role === ChatCompletionRequestMessageRoleEnum.Assistant
+ ) {
+ const value: AIChatItemValueItemType[] = [];
+
+ if (typeof item.reasoning_text === 'string') {
+ value.push({
+ type: ChatItemValueTypeEnum.reasoning,
+ reasoning: {
+ content: item.reasoning_text
+ }
+ });
+ }
+ if (item.tool_calls && reserveTool) {
+ // save tool calls
+ const toolCalls = item.tool_calls as ChatCompletionMessageToolCall[];
+ value.push({
+ //@ts-ignore
+ type: ChatItemValueTypeEnum.tool,
+ tools: toolCalls.map((tool) => {
+ let toolResponse =
+ messages.find(
+ (msg) =>
+ msg.role === ChatCompletionRequestMessageRoleEnum.Tool &&
+ msg.tool_call_id === tool.id
+ )?.content || '';
+ toolResponse =
+ typeof toolResponse === 'string' ? toolResponse : JSON.stringify(toolResponse);
+
+ return {
+ id: tool.id,
+ toolName: tool.toolName || '',
+ toolAvatar: tool.toolAvatar || '',
+ functionName: tool.function.name,
+ params: tool.function.arguments,
+ response: toolResponse as string
+ };
+ })
+ });
+ }
+ if (item.function_call && reserveTool) {
+ const functionCall = item.function_call as ChatCompletionMessageFunctionCall;
+ const functionResponse = messages.find(
+ (msg) =>
+ msg.role === ChatCompletionRequestMessageRoleEnum.Function &&
+ msg.name === item.function_call?.name
+ ) as ChatCompletionFunctionMessageParam;
+
+ if (functionResponse) {
+ value.push({
+ //@ts-ignore
+ type: ChatItemValueTypeEnum.tool,
+ tools: [
+ {
+ id: functionCall.id || '',
+ toolName: functionCall.toolName || '',
+ toolAvatar: functionCall.toolAvatar || '',
+ functionName: functionCall.name,
+ params: functionCall.arguments,
+ response: functionResponse.content || ''
+ }
+ ]
+ });
+ }
+ }
+ if (item.interactive) {
+ value.push({
+ //@ts-ignore
+ type: ChatItemValueTypeEnum.interactive,
+ interactive: item.interactive
+ });
+ }
+ if (typeof item.content === 'string') {
+ const lastValue = value[value.length - 1];
+ if (lastValue && lastValue.type === ChatItemValueTypeEnum.text && lastValue.text) {
+ lastValue.text.content += item.content;
+ } else {
+ value.push({
+ type: ChatItemValueTypeEnum.text,
+ text: {
+ content: item.content
+ }
+ });
+ }
+ }
+
+ return value;
}
- }
+
+ return [];
+ })();
return {
dataId: item.dataId,
diff --git a/packages/global/core/chat/type.d.ts b/packages/global/core/chat/type.d.ts
index 4e010c68f..8837075e8 100644
--- a/packages/global/core/chat/type.d.ts
+++ b/packages/global/core/chat/type.d.ts
@@ -77,6 +77,7 @@ export type AIChatItemValueItemType = {
| ChatItemValueTypeEnum.reasoning
| ChatItemValueTypeEnum.tool
| ChatItemValueTypeEnum.interactive;
+
text?: {
content: string;
};
diff --git a/packages/global/core/workflow/template/system/aiChat/index.ts b/packages/global/core/workflow/template/system/aiChat/index.ts
index 8d6aa6cd0..db5df3d6d 100644
--- a/packages/global/core/workflow/template/system/aiChat/index.ts
+++ b/packages/global/core/workflow/template/system/aiChat/index.ts
@@ -55,7 +55,7 @@ export const AiChatModule: FlowNodeTemplateType = {
showStatus: true,
isTool: true,
courseUrl: '/docs/guide/workbench/workflow/ai_chat/',
- version: '4813',
+ version: '490',
inputs: [
Input_Template_SettingAiModel,
// --- settings modal
diff --git a/packages/global/core/workflow/template/system/tools.ts b/packages/global/core/workflow/template/system/tools.ts
index 8ef75d9f5..672deaffa 100644
--- a/packages/global/core/workflow/template/system/tools.ts
+++ b/packages/global/core/workflow/template/system/tools.ts
@@ -58,6 +58,13 @@ export const ToolModule: FlowNodeTemplateType = {
valueType: WorkflowIOValueTypeEnum.boolean,
value: true
},
+ {
+ key: NodeInputKeyEnum.aiChatReasoning,
+ renderTypeList: [FlowNodeInputTypeEnum.hidden],
+ label: '',
+ valueType: WorkflowIOValueTypeEnum.boolean,
+ value: true
+ },
{
key: NodeInputKeyEnum.aiChatTopP,
renderTypeList: [FlowNodeInputTypeEnum.hidden],
diff --git a/packages/service/common/file/read/utils.ts b/packages/service/common/file/read/utils.ts
index 406c9f32f..2c943c6b1 100644
--- a/packages/service/common/file/read/utils.ts
+++ b/packages/service/common/file/read/utils.ts
@@ -245,7 +245,7 @@ export const readRawContentByFileBuffer = async ({
if (result_data.data.status === 'success') {
const result = result_data.data.result.pages
.map((page) => page.md)
- .join('\n')
+ .join('')
// Do some post-processing
.replace(/\\[\(\)]/g, '$')
.replace(/\\[\[\]]/g, '$$')
diff --git a/packages/service/core/ai/config/provider/Moonshot.json b/packages/service/core/ai/config/provider/Moonshot.json
index 796c529b8..f33b09ebe 100644
--- a/packages/service/core/ai/config/provider/Moonshot.json
+++ b/packages/service/core/ai/config/provider/Moonshot.json
@@ -75,6 +75,81 @@
"showTopP": true,
"showStopSign": true,
"responseFormatList": ["text", "json_object"]
+ },
+ {
+ "model": "moonshot-v1-8k-vision-preview",
+ "name": "moonshot-v1-8k-vision-preview",
+ "maxContext": 8000,
+ "maxResponse": 4000,
+ "quoteMaxToken": 6000,
+ "maxTemperature": 1,
+ "vision": true,
+ "toolChoice": true,
+ "functionCall": false,
+ "defaultSystemChatPrompt": "",
+ "datasetProcess": true,
+ "usedInClassify": true,
+ "customCQPrompt": "",
+ "usedInExtractFields": true,
+ "usedInQueryExtension": true,
+ "customExtractPrompt": "",
+ "usedInToolCall": true,
+ "defaultConfig": {},
+ "fieldMap": {},
+ "type": "llm",
+ "showTopP": true,
+ "showStopSign": true,
+ "responseFormatList": ["text", "json_object"]
+ },
+ {
+ "model": "moonshot-v1-32k-vision-preview",
+ "name": "moonshot-v1-32k-vision-preview",
+ "maxContext": 32000,
+ "maxResponse": 4000,
+ "quoteMaxToken": 32000,
+ "maxTemperature": 1,
+ "vision": true,
+ "toolChoice": true,
+ "functionCall": false,
+ "defaultSystemChatPrompt": "",
+ "datasetProcess": true,
+ "usedInClassify": true,
+ "customCQPrompt": "",
+ "usedInExtractFields": true,
+ "usedInQueryExtension": true,
+ "customExtractPrompt": "",
+ "usedInToolCall": true,
+ "defaultConfig": {},
+ "fieldMap": {},
+ "type": "llm",
+ "showTopP": true,
+ "showStopSign": true,
+ "responseFormatList": ["text", "json_object"]
+ },
+ {
+ "model": "moonshot-v1-128k-vision-preview",
+ "name": "moonshot-v1-128k-vision-preview",
+ "maxContext": 128000,
+ "maxResponse": 4000,
+ "quoteMaxToken": 60000,
+ "maxTemperature": 1,
+ "vision": true,
+ "toolChoice": true,
+ "functionCall": false,
+ "defaultSystemChatPrompt": "",
+ "datasetProcess": true,
+ "usedInClassify": true,
+ "customCQPrompt": "",
+ "usedInExtractFields": true,
+ "usedInQueryExtension": true,
+ "customExtractPrompt": "",
+ "usedInToolCall": true,
+ "defaultConfig": {},
+ "fieldMap": {},
+ "type": "llm",
+ "showTopP": true,
+ "showStopSign": true,
+ "responseFormatList": ["text", "json_object"]
}
]
}
diff --git a/packages/service/core/app/templates/templateSchema.ts b/packages/service/core/app/templates/templateSchema.ts
index 2485f535f..826437412 100644
--- a/packages/service/core/app/templates/templateSchema.ts
+++ b/packages/service/core/app/templates/templateSchema.ts
@@ -9,41 +9,23 @@ const AppTemplateSchema = new Schema({
type: String,
required: true
},
- name: {
- type: String
- },
- intro: {
- type: String
- },
- avatar: {
- type: String
- },
- author: {
- type: String
- },
+ name: String,
+ intro: String,
+ avatar: String,
+ author: String,
tags: {
type: [String],
default: undefined
},
- type: {
- type: String
- },
- isActive: {
- type: Boolean
- },
- userGuide: {
- type: Object
- },
- isQuickTemplate: {
- type: Boolean
- },
+ type: String,
+ isActive: Boolean,
+ userGuide: Object,
+ isQuickTemplate: Boolean,
order: {
type: Number,
default: -1
},
- workflow: {
- type: Object
- }
+ workflow: Object
});
AppTemplateSchema.index({ templateId: 1 });
diff --git a/packages/service/core/workflow/dispatch/agent/runTool/index.ts b/packages/service/core/workflow/dispatch/agent/runTool/index.ts
index d0f99583f..b10b57eac 100644
--- a/packages/service/core/workflow/dispatch/agent/runTool/index.ts
+++ b/packages/service/core/workflow/dispatch/agent/runTool/index.ts
@@ -55,7 +55,8 @@ export const dispatchRunTools = async (props: DispatchToolModuleProps): Promise<
userChatInput,
history = 6,
fileUrlList: fileLinks,
- aiChatVision
+ aiChatVision,
+ aiChatReasoning
}
} = props;
@@ -63,6 +64,9 @@ export const dispatchRunTools = async (props: DispatchToolModuleProps): Promise<
const useVision = aiChatVision && toolModel.vision;
const chatHistories = getHistories(history, histories);
+ props.params.aiChatVision = aiChatVision && toolModel.vision;
+ props.params.aiChatReasoning = aiChatReasoning && toolModel.reasoning;
+
const toolNodeIds = filterToolNodeIdByEdges({ nodeId, edges: runtimeEdges });
// Gets the module to which the tool is connected
diff --git a/packages/service/core/workflow/dispatch/agent/runTool/promptCall.ts b/packages/service/core/workflow/dispatch/agent/runTool/promptCall.ts
index 58e95a059..c5c0cb4bd 100644
--- a/packages/service/core/workflow/dispatch/agent/runTool/promptCall.ts
+++ b/packages/service/core/workflow/dispatch/agent/runTool/promptCall.ts
@@ -24,7 +24,12 @@ import {
import { AIChatItemType } from '@fastgpt/global/core/chat/type';
import { GPTMessages2Chats } from '@fastgpt/global/core/chat/adapt';
import { formatToolResponse, initToolCallEdges, initToolNodes } from './utils';
-import { computedMaxToken, llmCompletionsBodyFormat } from '../../../../ai/utils';
+import {
+ computedMaxToken,
+ llmCompletionsBodyFormat,
+ parseReasoningContent,
+ parseReasoningStreamContent
+} from '../../../../ai/utils';
import { WorkflowResponseType } from '../../type';
import { toolValueTypeList } from '@fastgpt/global/core/workflow/constants';
import { WorkflowInteractiveResponseType } from '@fastgpt/global/core/workflow/template/system/interactive/type';
@@ -58,6 +63,7 @@ export const runToolWithPromptCall = async (
temperature,
maxToken,
aiChatVision,
+ aiChatReasoning,
aiChatTopP,
aiChatStopSign,
aiChatResponseFormat,
@@ -216,7 +222,7 @@ export const runToolWithPromptCall = async (
const [requestMessages] = await Promise.all([
loadRequestMessages({
messages: filterMessages,
- useVision: toolModel.vision && aiChatVision,
+ useVision: aiChatVision,
origin: requestOrigin
})
]);
@@ -251,22 +257,46 @@ export const runToolWithPromptCall = async (
}
});
- const answer = await (async () => {
+ const { answer, reasoning } = await (async () => {
if (res && isStreamResponse) {
- const { answer } = await streamResponse({
+ const { answer, reasoning } = await streamResponse({
res,
toolNodes,
stream: aiResponse,
- workflowStreamResponse
+ workflowStreamResponse,
+ aiChatReasoning
});
- return answer;
+ return { answer, reasoning };
} else {
- const result = aiResponse as ChatCompletion;
+ const content = aiResponse.choices?.[0]?.message?.content || '';
+ const reasoningContent: string = aiResponse.choices?.[0]?.message?.reasoning_content || '';
- return result.choices?.[0]?.message?.content || '';
+ // API already parse reasoning content
+ if (reasoningContent || !aiChatReasoning) {
+ return {
+ answer: content,
+ reasoning: reasoningContent
+ };
+ }
+
+ const [think, answer] = parseReasoningContent(content);
+ return {
+ answer,
+ reasoning: think
+ };
}
})();
+
+ if (stream && !isStreamResponse && aiChatReasoning && reasoning) {
+ workflowStreamResponse?.({
+ event: SseResponseEventEnum.fastAnswer,
+ data: textAdaptGptResponse({
+ reasoning_content: reasoning
+ })
+ });
+ }
+
const { answer: replaceAnswer, toolJson } = parseAnswer(answer);
if (!answer && !toolJson) {
return Promise.reject(getEmptyResponseTip());
@@ -294,11 +324,16 @@ export const runToolWithPromptCall = async (
}
// No tool is invoked, indicating that the process is over
- const gptAssistantResponse: ChatCompletionAssistantMessageParam = {
+ const gptAssistantResponse: ChatCompletionMessageParam = {
role: ChatCompletionRequestMessageRoleEnum.Assistant,
- content: replaceAnswer
+ content: replaceAnswer,
+ reasoning_text: reasoning
};
- const completeMessages = filterMessages.concat(gptAssistantResponse);
+ const completeMessages = filterMessages.concat({
+ ...gptAssistantResponse,
+ reasoning_text: undefined
+ });
+
const inputTokens = await countGptMessagesTokens(requestMessages);
const outputTokens = await countGptMessagesTokens([gptAssistantResponse]);
@@ -379,9 +414,10 @@ export const runToolWithPromptCall = async (
})();
// 合并工具调用的结果，使用 functionCall 格式存储。
- const assistantToolMsgParams: ChatCompletionAssistantMessageParam = {
+ const assistantToolMsgParams: ChatCompletionMessageParam = {
role: ChatCompletionRequestMessageRoleEnum.Assistant,
- function_call: toolJson
+ function_call: toolJson,
+ reasoning_text: reasoning
};
// Only toolCall tokens are counted here, Tool response tokens count towards the next reply
@@ -502,12 +538,14 @@ ANSWER: `;
async function streamResponse({
res,
stream,
- workflowStreamResponse
+ workflowStreamResponse,
+ aiChatReasoning
}: {
res: NextApiResponse;
toolNodes: ToolNodeItemType[];
stream: StreamChatType;
workflowStreamResponse?: WorkflowResponseType;
+ aiChatReasoning?: boolean;
}) {
const write = responseWriteController({
res,
@@ -515,7 +553,9 @@ async function streamResponse({
});
let startResponseWrite = false;
- let textAnswer = '';
+ let answer = '';
+ let reasoning = '';
+ const { parsePart, getStartTagBuffer } = parseReasoningStreamContent();
for await (const part of stream) {
if (res.closed) {
@@ -523,13 +563,21 @@ async function streamResponse({
break;
}
- const responseChoice = part.choices?.[0]?.delta;
- // console.log(responseChoice, '---===');
+ const [reasoningContent, content] = parsePart(part, aiChatReasoning);
+ answer += content;
+ reasoning += reasoningContent;
- if (responseChoice?.content) {
- const content = responseChoice?.content || '';
- textAnswer += content;
+ if (aiChatReasoning && reasoningContent) {
+ workflowStreamResponse?.({
+ write,
+ event: SseResponseEventEnum.answer,
+ data: textAdaptGptResponse({
+ reasoning_content: reasoningContent
+ })
+ });
+ }
+ if (content) {
if (startResponseWrite) {
workflowStreamResponse?.({
write,
@@ -538,18 +586,20 @@ async function streamResponse({
text: content
})
});
- } else if (textAnswer.length >= 3) {
- textAnswer = textAnswer.trim();
- if (textAnswer.startsWith('0')) {
+ } else if (answer.length >= 3) {
+ answer = answer.trimStart();
+ if (/0(:|：)/.test(answer)) {
startResponseWrite = true;
+
// find first : index
- const firstIndex = textAnswer.indexOf(':');
- textAnswer = textAnswer.substring(firstIndex + 1).trim();
+ const firstIndex =
+ answer.indexOf('0:') !== -1 ? answer.indexOf('0:') : answer.indexOf('0：');
+ answer = answer.substring(firstIndex + 2).trim();
workflowStreamResponse?.({
write,
event: SseResponseEventEnum.answer,
data: textAdaptGptResponse({
- text: textAnswer
+ text: answer
})
});
}
@@ -557,7 +607,23 @@ async function streamResponse({
}
}
- return { answer: textAnswer.trim() };
+ if (answer === '') {
+ answer = getStartTagBuffer();
+ if (/0(:|：)/.test(answer)) {
+ // find first : index
+ const firstIndex = answer.indexOf('0:') !== -1 ? answer.indexOf('0:') : answer.indexOf('0：');
+ answer = answer.substring(firstIndex + 2).trim();
+ workflowStreamResponse?.({
+ write,
+ event: SseResponseEventEnum.answer,
+ data: textAdaptGptResponse({
+ text: answer
+ })
+ });
+ }
+ }
+
+ return { answer, reasoning };
}
const parseAnswer = (
@@ -568,8 +634,7 @@ const parseAnswer = (
} => {
str = str.trim();
// 首先，使用正则表达式提取TOOL_ID和TOOL_ARGUMENTS
- const prefixReg = /^1(:|：)/;
- const answerPrefixReg = /^0(:|：)/;
+ const prefixReg = /1(:|：)/;
if (prefixReg.test(str)) {
const toolString = sliceJsonStr(str);
@@ -585,13 +650,21 @@ const parseAnswer = (
}
};
} catch (error) {
- return {
- answer: ERROR_TEXT
- };
+ if (/^1(:|：)/.test(str)) {
+ return {
+ answer: ERROR_TEXT
+ };
+ } else {
+ return {
+ answer: str
+ };
+ }
}
} else {
+ const firstIndex = str.indexOf('0:') !== -1 ? str.indexOf('0:') : str.indexOf('0：');
+ const answer = str.substring(firstIndex + 2).trim();
return {
- answer: str.replace(answerPrefixReg, '')
+ answer
};
}
};
diff --git a/packages/service/core/workflow/dispatch/agent/runTool/type.d.ts b/packages/service/core/workflow/dispatch/agent/runTool/type.d.ts
index 64ecd79fc..61cb6b217 100644
--- a/packages/service/core/workflow/dispatch/agent/runTool/type.d.ts
+++ b/packages/service/core/workflow/dispatch/agent/runTool/type.d.ts
@@ -22,6 +22,7 @@ export type DispatchToolModuleProps = ModuleDispatchProps<{
[NodeInputKeyEnum.aiChatTemperature]: number;
[NodeInputKeyEnum.aiChatMaxToken]: number;
[NodeInputKeyEnum.aiChatVision]?: boolean;
+ [NodeInputKeyEnum.aiChatReasoning]?: boolean;
[NodeInputKeyEnum.aiChatTopP]?: number;
[NodeInputKeyEnum.aiChatStopSign]?: string;
[NodeInputKeyEnum.aiChatResponseFormat]?: string;
diff --git a/packages/service/core/workflow/dispatch/chat/oneapi.ts b/packages/service/core/workflow/dispatch/chat/oneapi.ts
index 41df8e31d..07035a826 100644
--- a/packages/service/core/workflow/dispatch/chat/oneapi.ts
+++ b/packages/service/core/workflow/dispatch/chat/oneapi.ts
@@ -563,6 +563,15 @@ async function streamResponse({
// if answer is empty, try to get value from startTagBuffer. (Cause: The response content is too short to exceed the minimum parse length)
if (answer === '') {
answer = getStartTagBuffer();
+ if (isResponseAnswerText && answer) {
+ workflowStreamResponse?.({
+ write,
+ event: SseResponseEventEnum.answer,
+ data: textAdaptGptResponse({
+ text: answer
+ })
+ });
+ }
}
return { answer, reasoning };
diff --git a/packages/service/worker/readFile/index.ts b/packages/service/worker/readFile/index.ts
index 625191a39..45092ed72 100644
--- a/packages/service/worker/readFile/index.ts
+++ b/packages/service/worker/readFile/index.ts
@@ -27,7 +27,9 @@ parentPort?.on('message', async (props: ReadRawTextProps