[Bug] o1-mini/o1-preview 返回 504

[ihnfsax]

📦 部署方式

Vercel

📌 软件版本

cda4494

💻 系统环境

Windows

📌 系统版本

10

🌐 浏览器

Edge

📌 浏览器版本

131.0.2875.0

🐛 问题描述

使用 Vercel 部署的服务，过去访问 o1-mini/o1-preview 没有问题。今天出现 504 错误，其他模型能够正常使用。

我使用的是 OpenAI 中转服务，在中转服务上能看到接收到请求和 token 被消耗。 Vercel 日志：

重新部署服务，以及调整 Function Max Duration 后重新部署都无法解决问题。

📷 复现步骤

No response

🚦 期望结果

No response

📝 补充信息

No response

[Issues-translate-bot]

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

---

Title: [Bug] o1-mini/o1-preview returns 504

📦 Deployment method

Vercel

📌 Software version

cda4494

💻 System environment

Windows

📌 System version

10

🌐 Browser

Edge

📌 Browser version

131.0.2875.0

🐛 Problem description

Using the service deployed by Vercel, there was no problem accessing o1-mini/o1-preview in the past. Got a 504 error today, other models are working fine.

I am using the OpenAI transit service. On the transit service, I can see that requests are received and tokens are consumed. Vercel log:

Redeploying the service, and adjusting the Function Max Duration and then redeploying did not solve the problem.

📷 Steps to reproduce

No response

🚦 Expected results

No response

📝 Supplementary information

No response

[ihnfsax]

也不是每一次请求中转服务都能记录到。记录到的从 vercel 发起请求，到中转服务收到模型输出，一般间隔 40~60s。但我把 Function Max Duration 设置为 60s 没有作用。

[Issues-translate-bot]

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

---

Not every request to the transit service can be recorded. It is recorded that the interval from when vercel initiates a request to when the transfer service receives the model output is generally 40~60s. But I set the Function Max Duration to 60s to no avail.