[Bug] 生成标题永远使用的model是gpt-4o-mini

[rufeng-h]

📦 部署方式

Docker

📌 软件版本

2.15.3

💻 系统环境

macOS

📌 系统版本

15.0

🌐 浏览器

Chrome

📌 浏览器版本

129.0.6668.60

🐛 问题描述

使用本地的ollama服务，CUSTOM_MODELS=-all,qwen2.5:7b-instruct,DEFAULT_MODEL=qwen2.5:7b-instruct，总结对话生成标题时model仍为gpt-4o-mini，生成的标题为后台错误提示json串。

📷 复现步骤

如上

🚦 期望结果

使用当前选择模型生成标题

📝 补充信息

无

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

Title: [Bug] The model always used to generate titles is gpt-4o-mini

📦 Deployment method

Docker

📌 Software version

2.15.3

💻 System environment

macOS

📌 System version

15.0

🌐 Browser

Chrome

📌 Browser version

129.0.6668.60

🐛 Problem description

Use the local ollama service, CUSTOM_MODELS=-all,qwen2.5:7b-instruct,DEFAULT_MODEL=qwen2.5:7b-instruct. When summarizing the conversation to generate the title, the model is still gpt-4o-mini, and the generated title is a background error prompt. json string.

📷 Steps to reproduce

As above

🚦 Expected results

Generate title using currently selected model

📝 Supplementary information

None

[QAbot-zh]

改下这里的设置