You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
How to use it when there is only one model locally? #202
My current issue is that my graphics memory is only sufficient to deploy one model locally, but the config requires two. How should I configure it to use one locally and one online? Because I see that only one api_base can be set in the config. Or how can I use two models with different api_bases?
You can use a single model by just putting the weight to 1.0. It is currently not possible to use different api_base but you can use an existing proxy or router like OptiLLM - https://github.com/codelion/optillm to router between models as needed with the same url.
Activity
codelion commentedon Aug 13, 2025
You can use a single model by just putting the weight to 1.0. It is currently not possible to use different api_base but you can use an existing proxy or router like OptiLLM - https://github.com/codelion/optillm to router between models as needed with the same url.