[Bug] 使用claude thinking时会出现max token小于thinking budgets

[fzlzjerry]

📦 Platform

Vercel

📦 Deploymenet mode

server db(lobe-chat-database image)

📌 Version

latest

💻 Operating System

macOS

🌐 Browser

Other

🐛 Bug Description

当将claude3.7的thinking budgets设置到64k的时候,会出现以下问题,但是我并没有开启单词回复最大用量的限制

{
  "error": {
    "type": "invalid_request_error",
    "message": "`max_tokens` must be greater than `thinking.budget_tokens`. Please consult our documentation at https://****/en/docs/build-with-claude/extended-thinking (request id: 20250312105351598489705tKNBl9ln)"
  },
  "endpoint": "https://api.ep****ne.chat/",
  "provider": "anthropic"
}

📷 Recurrence Steps

No response

🚦 Expected Behavior

No response

📝 Additional Information

No response

[canisminor1990]

👀 @fzlzjerry

Thank you for raising an issue. We will investigate into the matter and get back to you as soon as possible.
Please make sure you have given us as much context as possible.
非常感谢您提交 issue。我们会尽快调查此事，并尽快回复您。 请确保您已经提供了尽可能多的背景信息。

[dosubot]

The error you're encountering is due to the Claude API's token limit. The API enforces a maximum of 64,000 tokens, calculated as 64 * 1000, not 65,536 tokens. When you set the thinking budgets to 64k as 65,536 tokens, it exceeds the API's maximum allowed number of output tokens, resulting in the error message "max_tokens must be greater than thinking.budget_tokens" ^[1].

To resolve this, ensure that the thinking budget does not exceed 64,000 tokens. Adjust your settings accordingly to stay within this limit.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}