Skip to main content
The Reflection API serves Reflection models over HTTPS. The OpenAI-compatible endpoint implements Chat Completions and Models, so code that uses them through the official OpenAI SDKs works after you change the base URL and API key. See OpenAI compatibility for what is supported.
To work with Reflection models from your terminal instead of writing code, try Mirror, Reflection’s coding agent. Other coding agents can use the API too: see the Mirror CLI quickstart, which also covers them.

Start here

Quickstart

Make your first request in a few minutes.

Authentication

Create an API key and send it with your requests.

Models

See which models you can use.

Guides

Text generation

Send a conversation and get a response.

Streaming

Receive a response as it is generated.

Reasoning

Control how much the model thinks before it answers.

Tool calling

Let the model call functions in your code.

Structured outputs

Get JSON that matches a schema.

Resources

OpenAI compatibility

Use the OpenAI SDKs, and see what differs.

Errors

Error codes, and when to retry.

Rate limits

How limits apply and how to handle them.

API reference

Every endpoint, parameter, and response field.
Last modified on October 16, 2026