Open Model Gateway v0.3.0 documentation
One gateway to every model your organisation uses.
Open Model Gateway is a self-hosted AI model gateway for one enterprise. People and applications call cloud and self-hosted models through one API with their own keys; administrators decide who may use which models, set budgets and limits, and see what everything costs. These docs are for the people who call models, the administrators who run the gateway, and the operators who host it.
Sections
Getting started
What Open Model Gateway is, the ideas it is built on, a quick tour of the dashboard and a local demo.
Using the gateway
API keys, calling models with the OpenAI and Anthropic SDKs, logs, usage and costs, files, batches, realtime audio, alerts and workspace settings.
Administration
First run, people and roles, SSO groups and SCIM, connections, models, routes and prices, catalogs, limits and budgets, alerts, key safety, audit and settings.
API reference
The inference endpoints and what each supports: chat completions, responses, messages, embeddings, images, audio, rerank, files, batches, realtime and models.
Self-hosting
The container, every configuration variable, the database and migrations, identity, provider credentials, file storage, monitoring, backups, upgrades and security.
Releases
What changed in each release. These docs describe v0.3.0 (not yet released).
Open Model Gateway is pre-1.0 development software. These docs describe v0.3.0, the next release, as it stands on the main branch. More about the project is on its website; the engineering notes stay in the repository.