FAQ
Frequent questions grouped by topic - access and authentication, models and protocols, billing and limits, errors, and tool integrations.
Pick the category that matches where you're stuck. Every page lays its questions and answers out in full, and the table of contents on the right jumps straight to one.
Browse by category
| Category | What it covers |
|---|---|
| Access & authentication | Creating an API key, keeping it safe, auth headers, base URLs, balance and spend limits |
| Models & protocols | Where model IDs come from, which model uses which endpoint, Responses support, SDK compatibility, streaming |
| Billing & limits | Where to see usage, why Settled reads 0, whether failed media tasks are charged, rate limits and duplicate throttling |
| Errors | What to do about 400 / 401 / 402 / 424 / 429 / 5xx, and which ones not to retry |
| Tool integrations | Common snags with Claude Code, Codex CLI, and the one-click CC Switch import |
Not answered here
This section covers the frequent questions only. The complete rules live on the reference pages:
- Quick start - from creating a key to your first call
- Protocols & endpoints - which endpoint each model uses
- List models - what
GET /v1/modelsreturns - Billing & reconciliation - pricing, settlement, and reservation release
- Rate limits & reliability - limits and retry guidance
- Error codes - the full status-code and envelope reference