Use frontier open LLMs like Kimi K2, DeepSeek V3.1, GLM 4.5 and more in VS Code with GitHub Copilot Chat powered by Hugging Face Inference Providers 🔥
- Install the HF Copilot Chat extension here.
- Open VS Code's chat interface.
- Click the model picker and click "Manage Models...".
- Select "Hugging Face" provider.
- Provide your Hugging Face Token, you can get one in your settings page. You only need to give it the inference.serverless permissions.
- Choose the models you want to add to the model picker. 🥳
Each model entry also offers cheapest and fastest mode for each model. fastest selects the provider with highest throughput and cheapest selects the provider with lowest price per output token.
- Access SoTA open-source LLMs with tool calling capabilities.
- Single API to switch between multiple providers: Cerebras, Cohere, Fireworks AI, Groq, HF Inference, Hyperbolic, Nebius, Novita, Nscale, SambaNova, Together AI, and more. See the full list of partners in the Inference Providers docs.
- Built for high availability (across providers) and low latency.
- Transparent pricing: what the provider charges is what you pay.
💡 The free Hugging Face user tier gives you a small amount of monthly inference credits to experiment. Upgrade to Hugging Face PRO or Enterprise for $2 in monthly credits plus pay-as-you-go access across all providers!
- VS Code 1.104.0 or higher.
- Hugging Face access token with
inference.serverlesspermissions.
Team and Enterprise organizations can centralize Inference Providers billing: every member keeps using their own access token, but the requests are charged to the organization instead of the individual account.
Enable it from the extension:
- Open the chat model picker → Manage Models… → Hugging Face.
- Choose Bill to an Organization and enter your organization name (e.g.
my-org).
Or set it directly in your VS Code settings:
The extension forwards this value as the X-HF-Bill-To header on inference requests.
Resource groups. Enterprise organizations with Resource Groups can attribute inference costs to a specific group by entering the resource group ID instead of the organization name. Your token must be a member of that resource group.
ℹ️ Your access token must belong to the organization. Usage is tracked on the organization's billing page, where admins can set a spending limit and disable specific providers. Team & Enterprise organizations also receive a pool of free monthly credits based on the number of seats.
git clone https://github.com/huggingface/huggingface-vscode-chat
cd huggingface-vscode-chat
npm install
npm run compilePress F5 to launch an Extension Development Host.
Common scripts:
- Build:
npm run compile - Watch:
npm run watch - Lint:
npm run lint - Format:
npm run format
- Inference Providers documentation: https://huggingface.co/docs/inference-providers/index
- VS Code Chat Provider API: https://code.visualstudio.com/api/extension-guides/ai/language-model-chat-provider
- Open issues: https://github.com/huggingface/huggingface-vscode-chat/issues
- License: MIT License Copyright (c) 2025 Hugging Face