Ali Khokhar
Ali Khokhar
I haven't been able to reproduce maybe you set the wrong directory?
I tried implementing a similar idea before, but reverted it due to bugs. Instead of your current design what you can do is. Convert MODEL_* vars into comma separated fallback...
Ok so perform the rebasing, I am considering adding this provider next
This should be fixed
I have launched a fix in https://github.com/Alishahryar1/free-claude-code/commit/847916b1d3d81a57e5b8e69adbb556b143ad22ed, retry now
> > I have launched a fix in [847916b](https://github.com/Alishahryar1/free-claude-code/commit/847916b1d3d81a57e5b8e69adbb556b143ad22ed), retry now > > Deepseek v4 flash works like a charm, but deepseek v4 pro doesn't What's wrong with pro?
This should be fixed now
Why not just raise to very high limits?
Input is exceeding the model's context length, you will need to increase it