LM Studio Model?

[dustinwloring1988]

Question

I saw this support LM Studio and I was interested in fine-tuning an open source model to be better with the tool calls from opencode and more efficient with them. A few question before I start and also seeing if anyone would be interested in getting a dataset going.

• Has anyone done this yet with good results?
• Is anyone interested in helping with this?
• Do you currently use a reliable small local model that work well?

[github-actions[bot]]

This issue might be a duplicate of existing issues. Please check:

• #4851: How can I use the local model of ollama?

Feel free to ignore if none of these address your specific case.

[JosXa]

Interesting. One thing you could train it on is the parallel tools calls: https://github.com/sst/opencode/blob/dev/AGENTS.md

On another note, I had to manually specify my LM Studio models like this in the OC config.json:

    "lmstudio": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "LM Studio (local)",
      "options": {
        "baseURL": "http://127.0.0.1:1234/v1"
      },
      "models": {
        "mistral/devstral-small-2-24b-instruct-2512": {
          "name": "DevStral Small 2"
        }
      }
    }

HTH