Any plan for support automatic reasoning models size - like which is actually size of the non-full-scaled model

[foreverpiano]

Any plan for support automatic reasoning models size - like which is actually size of the non-full-scaled model

[gaocegege]

Sounds reasonable. Which model are you using?

[foreverpiano]

Like DeepSeek? I heard that some APIs will sometimes use the Qwen distilled version. Can u detect that through benchmark scores?

[gaocegege]

Do you suggest adding models like DeepSeek: R1 Distill Qwen 32B?

[foreverpiano]

yes