Reasoning level for DeepSeek V4 0731 @ Freebuff

[gimlichael]

Hello,

Since i am unable to set reasoning effort in the otherwise great CLI - what is the reasoning effort for the new DeepSeek V4?
I would love to use it - but its only valuable on max reasoning effort.

Same for the remainder of the models - if we cannot set the effort - can you please provide the baked in value?

Thanks.

[TanakritBenz]

Looks like it's not max
https://github.com/CodebuffAI/freebuff/blob/main/cli/src/cli-args.ts#L113-L117

[gimlichael]

Good finding .. i assume this is common for all offerings then .. lite is equal to low reasoning efforts. Good to know. Thanks.

Freebuff always uses LITE mode

[gimlichael]

Now i am confused ..

[TanakritBenz]

Yeah me too. Just noticed that after I replied 🤷‍♂️
I guess we're back to square one lol

[gimlichael]

Haha .. yeah; but fingers crossed they keep the transparency - high is good for the "price" - and I am positively surprised how fast it is 💨