调用 modelscope inference API 时候 resp解析失败

[thatname]

应该是不同的OpenAI API 都有一些差异。除了modelscope，像openrouter，还有本地的 tabbyAPI 之类的都可以试验一下。
平台 win，我用这个bat脚本来调用，其它平台应该差不多

@echo off
set OPENAI_BASE_URL=https://api-inference.modelscope.cn/v1
set OPENAI_API_KEY=ms-xxxxx-xxxxx-xxxxxx
set OPENAI_MODEL=Qwen/Qwen3-Coder-480B-A35B-Instruct

REM Get the directory where this batch file is located
set "SCRIPT_DIR=%~dp0"

REM Run coro.exe from the target/release folder relative to this batch file
"%SCRIPT_DIR%target\release\coro.exe" %*

输出是这样的：
███
░░░███
░░░███
░░░███
███░
███░
███░
░░░

Tips for getting started:

1. Ask questions, edit files, or run commands.
2. Be specific for the best results.
3. /help for more information.

There is a project called TexVerse on the github, which is a 3D mesh dataset. fetch it, then write python code
to stat how many vertices in random selected meshes downloaded from the dataset

🌓 Searching for TexVerse dataset… (2s · ↑ 95 tokens · esc to interrupt)

╭──────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ Type your message or @path/to/file (Enter to send, Shift+Enter for new line) │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────╯

Error: LLM request failed: LLM error: API error: 500 - failed to deserialize api response: missing field
error at line 1 column 100

[thatname]

实际上有的时候第一次调用时成功的