[Feature request ] Add multi model provider support

[azpublic]

Add the ability to configure multiple model providers ( anthropic , z.ai, kimi , and openai/ollama compatible providers etc... )
That way you can select whatever provider you want for the task at hand and switch either on a workspace basis or a per chat basis.

I have implemented the mechanism in my fork and it works nicely. Feel free to check it out. Thought it would be nice to have.

[1code-async]

Task completed.

There is no PR to review here — issue #146 is a feature request with no associated pull request in this repo. The issue author mentions having an implementation in their personal fork but hasn't submitted a PR.

Summary of Issue #146:

• Request: Multi-model provider support (Anthropic, xAI, Kimi, OpenAI/Ollama-compatible)
• Scope: Provider selection per-workspace or per-chat
• Status: No PR submitted — just a feature request

Key concerns if this were to become a PR:

1. Architecture impact — This app is tightly coupled to @anthropic-ai/claude-code SDK. Adding arbitrary providers would require either abstracting the SDK layer or running a separate completion path, which is a significant architectural change.
2. Tool execution model — Claude Code's agentic tool-use loop (bash, file edits, etc.) is Claude-specific. Other models have different tool-calling formats/capabilities, so "just swapping the model" isn't straightforward.
3. Scope creep risk — Supporting "OpenAI/Ollama compatible providers" is an open-ended surface area with many edge cases (streaming formats, token limits, tool schemas).
4. Per-chat vs per-workspace — Reasonable UX, but needs clear schema changes (provider config in chats or projects table).

If the author wants to contribute, they should open a PR from their fork so it can be reviewed concretely.

---

View full conversation

Powered by 1Code