api调用如何上传文件

[ioer2008]

我在image_url 中传入网址和base64,都无法成功,麻烦在文档中说明如果需要文件上传,该怎么操作

[rrrrrredy]

I ran a small compatibility test for image/file upload behavior against the LongCat OpenAI-compatible API.

This is related to the question in this issue: whether image_url, base64 images, or file upload are supported.

Environment

• Date: 2026-04-29
• Model: LongCat-Flash-Chat
• Base URL: https://api.longcat.chat/openai/v1
• Endpoint: /chat/completions
• Client: Python 3.12 + httpx

Summary

From this small test, LongCat-Flash-Chat appears to be text-only. I could not find a working path for image or file input.

The main observation is that image_url payloads are accepted with HTTP 200, but the model response indicates that it cannot see or analyze images. This may confuse clients that rely only on status code to detect vision support.

Test matrix

Method	HTTP status	Actually works?	Notes
Text-only chat	200	Yes	Baseline works
image_url with remote URL	200	No	API accepts request, but model says it cannot see images
image_url with base64 data URI	200	No	Same behavior as remote URL
GET /openai/v1/files	404	No	Files endpoint not found
POST /openai/v1/files	404	No	File upload endpoint not found
multipart /chat/completions	415	No	Chat completions appears to accept JSON only

Reproduction outline

For image_url, I sent a chat completion request with content array containing:

[
  {"type": "text", "text": "What color is this image?"},
  {
    "type": "image_url",
    "image_url": {
      "url": "https://via.placeholder.com/1x1/ff0000/ff0000.png"
    }
  }
]

The request returned HTTP 200, but the model replied that it could not see images directly.

I observed the same behavior with a data:image/png;base64,... image URL.

Questions

1. Is LongCat-Flash-Chat officially text-only at the API level?
2. Should image_url content be rejected with a 400-style error for this model, rather than being silently accepted?
3. Should the README/API documentation explicitly state that image/file upload is currently unsupported?
4. Is there a different endpoint or model variant intended for image/file inputs?

If the expected behavior is text-only, I would be happy to submit a small documentation PR clarifying supported and unsupported input types.