FIRZAWDEVELOPER RELAY / 01

PUBLIC API · OPENAI-COMPATIBLE

One relay.
Every useful shape.

A clean, authenticated gateway for coding clients, agents, and production workflows. Point an OpenAI-compatible client at the base URL, choose a model from the live catalog, and keep your prompts in control.

Start in 60 seconds ↘ Browse models →

Your key stays in your client. The relay never logs prompt or response content.

request.previewPOST
01POST https://api.firzaw.space/v1/chat/completions
02Authorization: Bearer $FIRZAW_API_KEY
03{
04"model": "cx/gpt-5.6-sol",
05"messages": [{ "role": "user" }]
06}
200 · stream ready~ 240ms
01Bearer authOne key, protected routes
02130+ modelsLive catalog, filtered locally
03OpenAI shapesChat Completions + Responses
04Anthropic bridgeMessages for compatible clients

01 / QUICKSTART

Make the first call
without guesswork.

Use the base URL below, set your key as an environment variable, and start with a model that is live right now. All examples use placeholders only.

A
Set the connectionEnvironment variables
https://api.firzaw.space/v1
$FIRZAW_API_KEY

Use a key issued for this relay. Never paste a production key into a browser, public issue, or client-side code.

B
Pick a modelLive catalog below

The catalog is read from the same gateway used for completions. It is a route list, not a promise that every provider is available for every request.

Inspect model metadata →

02 / MODEL EXPLORER

Know what the relay
can dispatch today.

●
Checking relayReading sanitized health data…
Loading model catalog…

03 / REFERENCE

Copy the shape.
Keep the control.

These are the stable public routes. The relay forwards compatible requests, adds a small set of routing conveniences, and keeps the response close to the upstream contract.

curl · chat completions
GET/v1/models

Discover model IDs and the sanitized capability metadata returned by the gateway.

curl https://api.firzaw.space/v1/models
POST/v1/chat/completions

OpenAI Chat Completions shape, including streaming responses when stream: true.

messages · tools · temperature · stream
POST/v1/responses

Responses shape for clients that prefer the newer OpenAI request envelope.

input · instructions · reasoning
POST/v1/messages

Anthropic Messages bridge for compatible clients. Availability depends on the selected route and active provider.

messages · system · stream

04 / TRANSPARENCY

Know the shape
before you ship.

↗Status is liveThe page checks a sanitized local health endpoint and the current model catalog.

⌁Errors are usefulUnauthorized requests receive a clear 401; invalid JSON and upstream failures are not disguised as successful empty responses.

◌No prompt telemetryRequest events contain routing metadata only. Prompts, responses, hidden reasoning, and credentials are excluded.