← Back to Docs

Zed with Meridian

Configure Zed's AI assistant to use Meridian models for inline completions, chat, and code transformations.

Step 1 — Open Zed Settings

Open the command palette with Cmd+Shift+P (macOS) or Ctrl+Shift+P (Linux/Windows), then select zed: open settings.

Step 2 — Add Meridian Provider

Paste the following block into your settings.json file. This registers Meridian as an OpenAI-compatible provider and sets it as the default assistant model.

{
  "language_models": {
    "meridian": {
      "provider": "openai_compatible",
      "model": "meridian-pro",
      "api_url": "https://api.getnimbus.net/v1",
      "available_models": ["meridian-pro", "meridian-flash"],
      "max_tokens": 8192
    }
  },
  "assistant": {
    "default_model": {
      "provider": "meridian",
      "model": "meridian-pro"
    }
  }
}

Step 3 — Authenticate

Set your Meridian API key as an environment variable. Zed reads provider keys from your shell environment automatically.

export MERIDIAN_API_KEY="mk_live_..."

Restart Zed after setting the variable. You can find your API key in the Meridian Dashboard.

Step 4 — Use It

Open the assistant panel with Cmd+Shift+Enter or use inline transformations by selecting code and pressing Ctrl+Enter. The model selector in the bottom-right corner should now show meridian-pro.

Note: Zed caches model lists on startup. If Meridian models don't appear, run zed: reload extensions from the command palette.