Connect → GitHub Copilot in VS Code

Use AxForge with GitHub Copilot in VS Code

The chat in VS Code (GitHub Copilot's chat, built in) can use models from any OpenAI-compatible service through its Custom Endpoint provider. Add AxForge once and our models sit in its model picker, beside the others.

Set it up

1. Add a Custom Endpoint

   Press Ctrl+Shift+P (Cmd+Shift+P on a Mac) and run Chat: Manage Language Models. Click Add Models, then Custom Endpoint.

   

2. Name it and paste your key

   Type AxForge as the group name and press Enter. Then paste your key and press Enter. VS Code keeps the key in its own secret store and shows it as dots.

   

3. Choose Chat Completions

   

4. Add the models

   VS Code opens chatLanguageModels.json with a group for AxForge. Keep its apiKey line as it is (that is VS Code's pointer to your stored key) and make "models" this list, then save:

   "models": [
       { "id": "qwen3.8-27b-nvfp4", "name": "Qwen3.8 27B", "url": "https://api.axforge.ai/v1/chat/completions",
         "toolCalling": true, "vision": true, "maxInputTokens": 245760, "maxOutputTokens": 16384 },
       { "id": "glm-5.3-flash", "name": "GLM-5.3-Flash", "url": "https://api.axforge.ai/v1/chat/completions",
         "toolCalling": true, "vision": false, "maxInputTokens": 245760, "maxOutputTokens": 16384 },
       { "id": "mimo-v2.6-flash", "name": "MiMo-V2.6-Flash", "url": "https://api.axforge.ai/v1/chat/completions",
         "toolCalling": true, "vision": false, "maxInputTokens": 245760, "maxOutputTokens": 16384 }
   ]

   The url is the whole address, /chat/completions included. Input and output together make 262,144 tokens, what each of the three can read at once.

   

5. Pick the model in the chat

   Open the chat (Chat: Open Chat), click the model name under the message box and type Qwen. Our models are under Other Models.

   

6. Ask something

   The answer comes from your AxForge key. The name under the message box says which model answered.

   

The settings

Which model for what: qwen3.8-27b-nvfp4 is the everyday default and reads images, glm-5.3-flash thinks before it answers, mimo-v2.6-flash is the fastest.

VS Code may show Set BYOK utility models. That offers to use your own models also for its small background jobs, such as chat titles. It is not needed for the chat to work.

Check that it works

The picker shows Qwen3.8 27B and a question gets an answer. The call also shows in the console under API Keys → Recent calls within a minute.

When it does not work

Test a call says in one click whether a problem is your key, your balance or us.