Codex OAuth image generation

Codex images. One bridge.

Use your Codex subscription through a CLI, private API, dashboard, or typed SDK. No OpenAI API key.

Install the CLI Explore the API

Authentication

Existing Codex and ChatGPT OAuth session

Default path

codex-responses with the Codex Responses backend

Deployment

Native binary or hardened multi-architecture container

Start where you are

Local workflow Generate from the command line Guided setup, diagnostics, generation, editing, presets, sessions, and verified output files. Read the CLI guide Service workflow Connect tools through one private API Bearer authentication, synchronous and streaming requests, durable jobs, metrics, and OpenAPI. Read the API contract

One contract across every surface

Imagegen Bridge normalizes requests, capability negotiation, metadata, errors, and provider attempts before they reach either supported Codex transport. Multiple requested images fan out concurrently by default; every concurrency and queue boundary can instead be set explicitly by the operator.

CLI and dashboard Interactive setup for people, complete JSON modes for automation.

HTTP and OpenAPI A private network service with bounded payloads, stable error codes, and actionable recovery suggestions.

Python and TypeScript Typed HTTP clients with streaming, deadlines, jobs, and diagnostics.

Rust Embed the complete runtime or register a compatible in-process provider.

The default Codex path

Your application
      |
      |  CLI, SDK, or authenticated HTTP
      v
Imagegen Bridge
      |
      +-- codex-responses     default and recommended
      |
      +-- codex-app-server    supported compatibility fallback
      |
      v
Codex / ChatGPT OAuth

codex-responses uses the Codex Responses backend and your existing Codex login. It does not read OPENAI_API_KEY. The codex-app-server transport remains available for compatibility, edits, references, and reusable Codex threads.

Continue with the guide you need