[Feature Request] 希望适配硅基流动平台上的deepseek-r1模型（思考内容可折叠）

[CykaZH]

🥰 需求描述

硅基流动上r1的思考内容放到reasoning_content里了，直接对话应该不显示思考内容

🧐 解决方案

输出reasoning_content的内容

📝 补充信息

No response

[Issues-translate-bot]

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

---

Title: null

🥰 Demand description

The thinking content of the R1 on the silicon -based flow is placed in the reasoning_content, and the direct conversation should not show the thinking content

🧐 Solution

Output the contents of Reasoning_Content

📝 Supplement information

_No Response

[CykaZH]

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

Title: null

🥰 Demand description

The thinking content of the R1 on the silicon -based flow is placed in the reasoning_content, and the direct conversation should not show the thinking content

🧐 Solution

Output the contents of Reasoning_Content

📝 Supplement information

_No Response

siliconflow (siliconflow.cn), not＂silicon -based flow＂

[jackalcooper]

@CykaZH 可以试试 https://nextchat-sf.vercel.app ，点击左下角 “SF“ 登录

[Issues-translate-bot]

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

---

@CykaZH can try https://nextchat-sf.vercel.app, click "SF" in the lower left corner

[jackalcooper]

不过现在显示有问题，如果reasoning包含 markdown 序号列表

[jackalcooper]

经过一番折腾，似乎必须把siliconflow加入成为独立的provider才能完善地兼容，因为自定义的模型会导致 provider 叫做 deepseek-ai/DeepSeek-R1 从而无法用 this.llm = new DeepSeekApi(); 把 api 对象正确设置成 deepseek 版本

[Issues-translate-bot]

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

---

After some tossing, it seems that Siliconflow must be added as an independent provider to be compatible, because the custom model can cause the Provider to be called Deepseek-AI/Deepseek-R1 and cannot usethis.llm = new deeepseekapi (). ;;Correctly set the API object to the DEEPSEEK version

[jackalcooper]

经过一番折腾，似乎必须把siliconflow加入成为独立的provider才能完善地兼容，因为自定义的模型会导致 provider 叫做 deepseek-ai/DeepSeek-R1 从而无法用 this.llm = new DeepSeekApi(); 把 api 对象正确设置成 deepseek 版本

但是如果把模型设置成 deepseek-reasoner，确实能够创建 DeepSeekApi()，而 SiliconFlow 的模型名字又不叫 deepseek-reasoner 而叫做 deepseek-ai/DeepSeek-R1

[Issues-translate-bot]

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

---

经过一番折腾，似乎必须把siliconflow加入成为独立的provider才能完善地兼容，因为自定义的模型会导致provider 叫做deepseek-ai/DeepSeek-R1 从而无法用`this.llm = new DeepSeekApi() ;

But if the model is set to Deepseek-Reasoner, it can indeed createDeepseekapi (), and SiliconFlow's model name is not called Deepseek-Reasoner and called Deepseek-AI/Deepseek-R1

[liaokaime]

Many providers, such as Azure and Amazon, have started supporting the R1 model. There should be a way to make all R1 models compatible.

[jackalcooper]

#6141 should fix this

[ChaoYan]

不过现在显示有问题，如果reasoning包含 markdown 序号列表

如果生成内容有公式可能也有显示问题

[Issues-translate-bot]

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

---

However, there is a problem with the display now, if reasoning contains the markdown serial number list

If there are formulas for generating content, there may be problems with displaying them.

[wl223600]

Many providers, such as Azure and Amazon, have started supporting the R1 model. There should be a way to make all R1 models compatible.

的确。百度和阿里云的deepseek-r1在网页端也都提供了思考过程，但是在NextChat里都未能显示。目前Merged的PR还是仅解决了单个平台的问题。