Hi @lofty elbow. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
#More BYOK options?
warm pecanBOT
Warp users have been actively requesting broader BYOK (Bring Your Own Key) support, including the ability to use "OpenAI-compatible" endpoints and other providers like AskSage. Many users have specifically asked for Warp to allow specifying any endpoint that implements the OpenAI API standard, which would enable compatibility with a wide range of local and remote LLMs (such as vLLM, Ollama, LM Studio, and others). This would make Warp more flexible and attractive for those who want to use self-hosted or alternative AI providers, not just OpenAI itself. The Warp team has acknowledged these requests and indicated that BYOK and local LLM support are on their roadmap, but there is no explicit confirmation or ETA for support of "OpenAI-compatible" endpoints or AskSage at this time.
For example, one highly upvoted suggestion states:
Warp should be able to accept any model based on OpenAI API standard. Allowing users to input the endpoint address and, using the API standard from OpenAI, interact with the LLM model. This will make the Warp terminal compatible with any LLM implementation. Be this vLLM, Dynamo, <name> or any other inference service local or remote.
Suggestion for Adding Support for Custom AI Model Configuration
The Warp team is tracking these feature requests and has stated that BYOK is actively being worked on, but details about which providers or compatibility layers will be supported are not yet available. If you want to see support for "OpenAI-compatible" endpoints or AskSage, it's recommended to upvote and comment on the relevant GitHub issues to show your interest and follow updates as they are posted by the Warp team.
In summary:
- BYOK is coming and is now available in some form for non-enterprise users.
- There is strong user demand for support of "OpenAI-compatible" endpoints and other providers.
- The Warp team has not yet confirmed support for AskSage or generic OpenAI-compatible endpoints, but they are aware of the request and it is under consideration.