关于接入本地模型，我的是mac然后lm studio，但是验证零个模型

[Elixk]

问题描述

111

相关上下文

No response

[PosenChen]

我想接入區域網路內LM studio部署的本地模型時也發生同樣的問題

[Ngchunwei]

Hi Qclaw Team,
I'm experiencing an issue where Qclaw cannot detect locally loaded models from LM Studio, even though the connection is successful.

Environment:

• Device: MacBook with Apple M1, 8GB RAM
• Qclaw version: 3.2905
• LM Studio version: LM Studio 0.4.9 (Build 1)
• Model: Qwen3.5-4B-GGUF (Q4_K_M)

Steps to reproduce:

1. Loaded Qwen3.5-4B in LM Studio — status shows READY
2. LM Studio Local Server is running at http://127.0.0.1:8000/
3. In Qclaw → 模型与API → 添加新供应商 → selected vLLM (本地)
4. Tried the following Base URLs:
   • http://127.0.0.1:8000/v1
   • http://127.0.0.1:8000/api/v1
   • http://127.0.0.1:8000/

Expected behavior:
Qclaw should detect and list the loaded model (qwen3.5-4b).
Actual behavior:
Qclaw shows "连接成功，已发现 0 个本地模型" (Connected successfully, 0 models found) for all URL variations.
Additional info:
I confirmed via LM Studio Developer Logs that the server IS receiving the GET /api/v1/models request and returning 2 models successfully:

Received request: GET to /api/v1/models
Returning 2 models from v1 API

This suggests Qclaw is connecting correctly but failing to parse the model list returned by LM Studio.
Please advise on the correct configuration, or if this is a compatibility issue with the latest LM Studio API format.
Thank you!

[gaoqi125]

同样的问题