Zed: using the VIDAI Control Plane as the backend¶

Zed's AI assistant supports custom OpenAI-compatible providers. Add one to your settings.json, point it at the control plane, and every assistant chat and inline transformation runs through us.

TL;DR¶

Open Zed → zed: open settings and add:

{
  "language_models": {
    "openai_compatible": {
      "vidai-control-plane": {
        "api_url": "https://your-vidai-server.example.com/v1",
        "api_key_env_var": "VIDAI_API_KEY",
        "available_models": [
          { "name": "gpt-4o-mini", "display_name": "GPT-4o mini (via VIDAI)", "max_tokens": 128000 }
        ]
      }
    }
  }
}

Then set the env var Zed reads:

export VIDAI_API_KEY=your-vidai-key

Open the assistant panel, pick GPT-4o mini (via VIDAI) as the model, and go.

Prerequisites¶

Zed installed.
Control plane base URL and an API key from API Keys.
A model registered on the Models page.

Add multiple models¶

Extend available_models with every model your team should be able to pick:

"available_models": [
  { "name": "gpt-4o-mini",       "display_name": "GPT-4o mini (via VIDAI)", "max_tokens": 128000 },
  { "name": "claude-sonnet-4.5", "display_name": "Claude Sonnet (via VIDAI)", "max_tokens": 200000 },
  { "name": "gemini-2.0-flash",  "display_name": "Gemini 2 Flash (via VIDAI)", "max_tokens": 1000000 }
]

Each appears in the assistant's model picker.

Inline transformations¶

The assistant: inline assist command (Cmd+Enter) uses whichever model is currently active. Once your VIDAI model is selected, inline edits flow through the control plane.

Verify it works¶

Open the assistant, select your VIDAI model, send:

reply with exactly: ok

A row appears on Request Logs.

If something's off¶

Raise an issue at github.com/vidaiUK/vidai-quickstart/issues with your settings.json provider block (masking the key) and the Zed version. We'll get it sorted.