AI inference through one clean API.

Start here

1. Get your CostIQ API key. The shared base URL is https://inference.costiq.xyz/v1 and the model alias is general-fast.
2. Run the four-step smoke test to verify authentication, model access, native Responses, and streaming.
3. Choose a production-qualified setup for Codex or OpenCode, or use an SDK example.
4. For another agent, check the compatibility contract before relying on it.

The API

CostIQ exposes a deliberately small public surface. Discover the models available to your key, then create native Responses or Chat Completions with or without streaming.

Your credentials

Every person, application, and environment should use a separate CostIQ key. Each credential can have its own expiry, model access, request rate, token rate, concurrency, and estimated-spend budget.

Make a request

With COSTIQ_BASE_URL and COSTIQ_API_KEY set in your shell:

curl "$COSTIQ_BASE_URL/responses" \
  -H "Authorization: Bearer $COSTIQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "general-fast",
    "input": "Explain this in one sentence.",
    "store": false
  }'

Where to go next