CLLMs for Copilot Chat

中文

Thanks

CLLMs began as a Qwen-focused adaptation of Vizards/deepseek-v4-for-copilot by Vizards, which pioneered the approach of plugging a BYOK model into the Copilot Chat picker via the native LanguageModelChatProvider API, and has since grown into a multi-provider extension for Qwen, z.ai (GLM), MiniMax, Xiaomi MiMo, Moonshot Kimi, and Tencent Hunyuan. Huge thanks to the original author — the provider pipeline, vision proxy, thinking-mode handling, and diagnostics here are deeply inspired by and built upon the generous foundation that Vizards created and shared with the community.

Getting Started

Prerequisites

• VS Code 1.116 or later. This extension relies on non-public Copilot Chat APIs that may break on newer VS Code versions — report an issue if you hit one.
• GitHub Copilot subscription (Free / Pro / Enterprise — the free tier works)
• An API key for at least one provider:
  • Qwen: a DashScope (Alibaba Cloud Model Studio) key from Model Studio, or any compatible token when using a custom cllms.baseUrl.
  • z.ai (GLM): a key from the z.ai API keys page.
  • MiniMax: a key from the MiniMax platform.
  • Xiaomi MiMo: a pay-as-you-go (sk-...) key from the Xiaomi MiMo open platform.
  • Moonshot (Kimi): a key from the Moonshot console (international; platform.moonshot.cn for mainland China).
  • Tencent Hunyuan (混元): a key from the Tencent Cloud Hunyuan console.

Installation

Install from the registry used by your editor:

1. Microsoft VS Code — install from VS Code Marketplace.

Usage

1. Run CLLMs: Set API Key from the Command Palette (Cmd+Shift+P) and pick a provider
2. Paste that provider's API key or compatible token (Qwen DashScope keys usually start with sk-)
3. Open Copilot Chat, click the model picker, pick a model
4. That's it — chat away

Models

Six providers ship out of the box. Each model carries its own API key and endpoint, so you can use Qwen, z.ai (GLM), MiniMax, Xiaomi MiMo, Moonshot Kimi, Tencent Hunyuan, or any combination at the same time from the Copilot model picker.

Qwen (DashScope)

z.ai (Zhipu GLM)

MiniMax

Xiaomi MiMo

Moonshot (Kimi)

Tencent Hunyuan (混元)

Model IDs are the official provider names and are fully configurable via cllms.modelIdOverrides / cllms.zai.modelIdOverrides / cllms.minimax.modelIdOverrides / cllms.xiaomi.modelIdOverrides / cllms.moonshot.modelIdOverrides / cllms.hunyuan.modelIdOverrides for third-party / self-hosted endpoints.

Adding a new model

Want to add your own model? See Adding a new model for a step-by-step guide.

Testing Status

💡 Help wanted! International endpoints share the same API surface as their domestic counterparts, so they should work out of the box — but they haven't been verified yet. If you have an international API key, please give it a try and report your results. If you'd like to contribute test tokens, reach out via GitHub Issues. Every bit of testing helps make these providers more reliable for everyone.

Settings

Thinking Effort is configured from Copilot Chat's model picker for each thinking-capable model.

Commands

Run these from the Command Palette (Cmd/Ctrl+Shift+P):

Example settings.json override for compatible API proxies:

{
  "cllms.modelIdOverrides": {
    "qwen3-coder-plus": "your-coder-model-id",
    "qwen-plus": "your-plus-model-id",
    "qwen3-max": "your-max-model-id",
    "qwen3-vl-plus": "your-vl-model-id"
  }
}

Using z.ai (Zhipu GLM)

z.ai is a first-class provider — no proxy or model-ID hacking required:

1. Run CLLMs: Set API Key and pick z.ai (Zhipu GLM). Get a key from the z.ai API keys page.
2. Open the Copilot Chat model picker — the GLM models appear alongside the Qwen ones.

GLM thinking is sent in z.ai's native format (thinking: { type: "enabled" }), tool calling works, and GLM-4.5V is used as a native vision model (images sent directly). If you have a GLM Coding Plan subscription, set cllms.zai.baseUrl to https://api.z.ai/api/coding/paas/v4.

Using MiniMax

MiniMax is also a first-class provider:

1. Run CLLMs: Set API Key and pick MiniMax. Get a key from the MiniMax platform.
2. Open the Copilot Chat model picker — MiniMax-M3 and MiniMax-M2.7 appear alongside the others.

MiniMax thinking is sent in its native format (thinking: { type: "adaptive" }) and reasoning is requested via reasoning_split: true so it streams cleanly through reasoning_content. Tool calling works; MiniMax-M3 is a native vision model (images sent directly), while MiniMax-M2.7 is text-only and image attachments use the vision proxy fallback. The default endpoint is the international https://api.minimax.io/v1 — set cllms.minimax.baseUrl to https://api.minimaxi.com/v1 for mainland China.

Using Xiaomi MiMo

Xiaomi MiMo is a first-class provider too:

1. Run CLLMs: Set API Key and pick Xiaomi MiMo. Create a pay-as-you-go (sk-...) key on the Xiaomi MiMo open platform console.
2. Open the Copilot Chat model picker — MiMo V2.5 Pro, MiMo V2.5 (Omni), and MiMo V2 Flash appear alongside the others.

MiMo is a hybrid-reasoning family: thinking is on by default and sent in the same format as GLM (thinking: { type: "enabled" | "disabled" }; MiMo doesn't support a thinking budget), with reasoning streamed through reasoning_content. Tool calling works, and the omni model MiMo V2.5 (Omni) accepts native image input while the Pro/Flash models fall back to the vision proxy. The default endpoint is the official open platform https://api.xiaomimimo.com/v1.

Note: a MiMo Token Plan subscription (tp-... key) uses a different, subscription-specific base URL and is restricted to coding tools — point cllms.xiaomi.baseUrl at the URL shown on your subscription page if you use one. Pay-as-you-go (sk-...) keys work with the default endpoint.

Using Moonshot (Kimi)

Moonshot Kimi is a first-class provider too:

1. Run CLLMs: Set API Key and pick Moonshot (Kimi). Create a key in the Moonshot console.
2. Open the Copilot Chat model picker — Kimi K2.6 and Kimi K2.5 appear alongside the others.

Kimi K2.6 / K2.5 are native-multimodal hybrid-reasoning models (256K context): thinking is on by default and sent in the GLM-style thinking: { type: "enabled" | "disabled" }, with reasoning streamed through reasoning_content. Tool calling works, and both models accept native image input. The default endpoint is the international https://api.moonshot.ai/v1 — set cllms.moonshot.baseUrl to https://api.moonshot.cn/v1 for mainland China.

Note: Moonshot keys are region-specific — an international (platform.moonshot.ai) key only works against api.moonshot.ai, and a mainland-China (platform.moonshot.cn) key only works against api.moonshot.cn. The legacy kimi-k2-* series (incl. kimi-k2-thinking) was retired on 2026-05-25; use K2.6 / K2.5.

Using Tencent Hunyuan (混元)

Tencent Hunyuan is a first-class provider too:

1. Run CLLMs: Set API Key and pick Tencent Hunyuan (混元). Get an API key from the Tencent Cloud Hunyuan console.
2. Open the Copilot Chat model picker — Tencent HY 2.0 Think, Hunyuan TurboS, Hunyuan T1, and Hunyuan A13B appear alongside the others.

Hunyuan uses the standard OpenAI-compatible Chat Completions API. HY 2.0 Think and T1 are deep-thinking models with thinking sent in GLM-style thinking: { type: "enabled" | "disabled" }; TurboS and A13B are fast instruct models. Tool calling works across all four models; there are no native vision models yet, so image attachments use the vision proxy fallback. The default endpoint is https://api.hunyuan.cloud.tencent.com/v1.

Other OpenAI-compatible providers

Beyond the six built-in providers, requests go through a standard OpenAI-compatible Chat Completions endpoint, so you can repoint any provider at a compatible service via its baseUrl and map IDs with modelIdOverrides. For example, to serve GLM through the Qwen slots instead:

{
  "cllms.baseUrl": "https://api.z.ai/api/paas/v4",
  "cllms.modelIdOverrides": {
    "qwen3-coder-plus": "glm-4.6"
  }
}

First-class, named entries for more Chinese providers are on the roadmap.

License

MIT — see NOTICE for attribution to the upstream project.