Skip to content

GitHub Copilot with Keln

Use Keln models in Copilot Chat in VS Code by adding Keln as a custom endpoint. The models then appear in the chat model picker.

What you need to start

  • VS Code 1.129 or later with Copilot Chat.
  • A Keln API key. Sign up at app.keln.ai, then open Keys → Create key.

On Copilot Business and Enterprise, an administrator can turn off the Bring Your Own Language Model Key in VS Code policy. If Add Models is greyed out, ask your administrator.

Connect Copilot Chat to Keln

1. Open the Language Models editor

Open the Chat view. In the model picker of the chat input, click Manage Language Models (the gear icon). You can also run Chat: Manage Language Models from the Command Palette.

2. Add a Custom Endpoint

Click Add Models and choose Custom Endpoint. VS Code asks for 3 values:

VS Code asks for Enter
Group name Keln
API key sk-keln-… your key
API type Chat Completions

The Add Models menu in the Language Models editor, with Custom Endpoint as the last entry

3. List the Keln models

VS Code opens chatLanguageModels.json with a new entry for Keln. Replace only the models list in that entry with the Keln models you want. Keep the other lines as VS Code wrote them. The apiKey line points to your saved key:

[
  {
    "name": "Keln",
    "vendor": "customendpoint",
    "apiKey": "${input:chat.lm.secret.…}",
    "apiType": "chat-completions",
    "models": [
      {
        "id": "zai-org/glm-5.3",
        "name": "zai-org/glm-5.3",
        "url": "https://api.keln.ai/v1/chat/completions",
        "toolCalling": true,
        "vision": false,
        "contextWindow": 1048576,
        "maxOutputTokens": 32768,
        "supportsReasoningEffort": ["low", "medium", "high"]
      },
      {
        "id": "moonshotai/kimi-k3",
        "name": "moonshotai/kimi-k3",
        "url": "https://api.keln.ai/v1/chat/completions",
        "toolCalling": true,
        "vision": true,
        "contextWindow": 1048576,
        "maxOutputTokens": 32768,
        "supportsReasoningEffort": ["low", "medium", "high"]
      }
    ]
  }
]

The supportsReasoningEffort line adds an Effort Level picker (Low, Medium, High) for the model. Without it, Keln runs the model at its deepest thinking level.

Save the file.

4. Pick a Keln model

Open the model picker in the chat input and choose a Keln model. If the models are missing, restart VS Code.

Before your first message, add credit under Billing → Add credit. Keln returns a 402 error if your balance is empty.

Models

Copilot Chat works with all Keln chat models. Add one models entry for each model you want. See Models & pricing for the full list, with each model's context window, image support and pricing.

Usage

Your Keln Usage page shows your spend, tokens, latency and error rate. Keln stores metadata only (ZDR).

The Keln Usage page with sample data: spend, requests, tokens, time to first token, error rate and a spend-by-day chart

What stays on GitHub

Keln serves the chat and agent requests for the models you added. Inline code suggestions, semantic search, chat titles and commit messages stay on your GitHub Copilot account. The setting Chat: Byok Utility Model Default chooses the model for VS Code's built-in utility tasks when your chat model is a Keln model.

Switch back

Pick a built-in model in the chat model picker. To remove Keln, open Manage Language Models and delete the Keln group.

Where to go next

Try Keln

Create a Keln account and API key to run open models in GitHub Copilot.

Get an API key Models & pricing ↗