Use Zed with GammaInfra

Zed is fast — Rust-based, GPU-accelerated, no Electron tax. Drop GammaInfra's smart router in via one settings.json change and you get per-completion cost visibility and automatic provider fallback across every major LLM, without losing Zed's speed advantage.

What changes with GammaInfra

Zed's AI assistant talks to OpenAI by default. Routing it through GammaInfra:

• Every assistant response carries X-GammaInfra-Cost-USD — visible via the dashboard rollups at dashboard.gammainfra.com.
• Provider rate-limits cascade through GammaInfra's fallback chain transparently. Zed sees a 200, your assistant conversation continues.
• You can pin specific models (anthropic/claude-opus-4-7, openai/gpt-5, etc.) through one configuration.

Setup

1. Get a GammaInfra API key

Sign up at gammainfra.com and verify your email.

2. Open Zed settings

Cmd/Ctrl + , opens settings (or use the command palette: zed: open settings).

3. Configure the openai provider

Zed's settings.json supports a language_models block:

{
  "language_models": {
    "openai": {
      "api_url": "https://api.gammainfra.com/v1",
      "available_models": [
        { "name": "gammainfra/auto",                  "max_tokens": 16000 },
        { "name": "anthropic/claude-opus-4-7",    "max_tokens": 16000 },
        { "name": "anthropic/claude-sonnet-4-6",  "max_tokens": 16000 },
        { "name": "openai/gpt-5",                 "max_tokens": 16000 },
        { "name": "deepseek/deepseek-v4-pro",     "max_tokens": 16000 }
      ]
    }
  }
}

4. Set your API key

Set OPENAI_API_KEY in your shell environment (Zed reads it for the openai provider) OR add it via Zed's secrets management:

export OPENAI_API_KEY="sk-gammainfra-..."

Restart Zed so it picks up the new env var.

5. Use the assistant panel

Cmd/Ctrl + Shift + Y opens the AI assistant panel. Pick a GammaInfra-routed model from the dropdown.

Verify it's working

Open the assistant panel, ask a question, then check the dashboard:

• Recent requests appear with the resolved model, latency, cost.
• Cost rollups by day show where Zed traffic lands.

Trade-offs

• Latency. ~10–50 ms overhead vs going direct. Zed's UI thread doesn't block on the LLM call, so this is invisible.
• Cost. 3% top-up fee (launch) / 5% standard. Pass-through provider rates on tokens — no markup.
• Privacy. Prompts and responses aren't logged by default. Privacy policy.

Ready to try it?

Get a GammaInfra API key →

$3 free trial credit on signup, $10 minimum top-up. Pass-through provider token rates plus 3% top-up fee during the launch window.