Bring your own inference to Warp

Warp now provides developers with more control over AI inference by allowing them to use their own models and provider accounts. Users on the free plan can bring their own API key (BYOK) for providers like OpenAI, Anthropic, and Google. The platform also supports custom inference endpoints compatible with the OpenAI Chat Completions API, enabling connections to model routers, gateways, or self-hosted setups. This update gives developers greater flexibility over model choice and cost, while future plans include a client-side harness for local models and support for the Agent Client Protocol.