Right now, Railway's AI features use a single model configuration with no flexibility. If you already have an OpenAI, Anthropic, or other LLM subscription, you can't use it — and you can't pick the model you prefer.

We're adding support for bringing your own API keys and native sign-ins so you can use the models you want across Railway's AI features. Each user in a workspace will be able to configure their own provider independently. What providers or models would you most want to use? Let us know!