Allow Applying Ki on any LLM

vicky

Users should be allowed to apply Ki as a config on any model.
Say the user wants to use o3, o3 pro or Claude 4.1 while retaining the research capacity of Ki, they should be able to do so.

So the user can use ki config on any model, they want giving them more access. This can be limited to Ultimate users only, but I think this would be a great feature to have

Thibaultmol

While I get the want for this, the problem seems to be that all kinds of models act differently when you give it a Ki style prompt. So they might have to design a prompt for each of the models.
But while having it as a 'experimental, use at your own risk' option, I think it might be awkward to use if it fails and such

RoxyRoxyRoxy

Hi, can I ask what the use case of this would be? What are you getting out of other models that you don't from Ki?

vicky

RoxyRoxyRoxy

So say like I want to use o3 pro because it thinks a lot more, or say grok 4, and I want it to do deep research similar to how Ki does, then this would be useful.
Especially once the Kagi assistant API comes out. A user using the API may want to use specifically o3 pro, or claude Opus 4.1 for eg while having the same deep tool research capability as Ki

Rehan

We are considering maybe having something like this down the line. The issue is that, as Thibaultmol pointed out, the pipeline is molded around the base models used in a way. But we're looking into it