Compatibility at a glance
| Capability | Scope |
|---|---|
| Clients | Codex App and Codex CLI; use the matching Console preset. |
| Protocol | Responses routing at https://api.superfastcode.com/v1. |
| Authentication | A Superfastcode Codex/OpenAI-platform key. The default preset preserves the existing local login file. |
| Streaming transport | Default HTTP preset and a separately generated WebSocket preset. |
| Models and rates | Use the current Codex catalog, selected key policy and account limits. |
| Other SDKs or agents | Not implied by the Codex channel. A compatible JSON shape is not an access entitlement. |
Connect and verify
- Get a matching key from Console → API Keys.
- Follow the public custom provider guide, or use the current generated preset for your client.
- Restart the client and complete the small smoke test.
- Check the matching request and charged usage in Console.
A successful health check confirms reachability; a model list confirms a catalog response. Neither establishes that your selected key can complete a request to your selected model. For tools or image functionality, run a separate representative task and inspect its actual result.
Billing and operational limits
Gateway requests use Superfastcode billing. Your ChatGPT subscription is not a Superfastcode credit balance. Input, output and cache rates, context rules, model-specific multipliers and optional modes can affect charges; read the model pricing notes along with your account plan.
Model availability, latency and feature support can vary with client version, channel, region and provider conditions. A configured model is not a reservation or uptime guarantee. Avoid simultaneous retry loops when the account reaches concurrency or usage limits.
Advanced client features may need a specific Console preset, a full restart and a new task. This page does not promise every App feature, extension or cloud workflow. Share the client version and a redacted failing request reference with support when a feature is uncertain.
Service identity and request data
Superfastcode is operated by SuperfastAI LLC and is independent of OpenAI. OpenAI product names identify the clients and protocols; they do not imply endorsement. Requests can pass through third-party providers and infrastructure.
Read the Privacy Policy before sending proprietary code or regulated information. A local client setting such as disable_response_storage is not a gateway-wide retention commitment. Questions: support@superfastai.org.
Sources checked October 16, 2026: Official OpenAI configuration documentation; Superfastcode’s current client presets, terms and channel catalog.