← All IntelClip / OtherDeep research fusion result: smart model as the outer wrapper
From The State of Model Routing — NVIDIA, Cognition, OpenRouter · ≈17:29
“we had the smart model be the wrapper model, be the outer model, and we got the best results from doing that.”
“it's fusion is not super well optimized for coding.”
“it might be that like a smaller model ends up being like a higher um efficiency per you know, fewer dollars um per like completed successfully completed task, but it's kind of early”
What’s in it
- Reveals early results on 'model fusion' combining smart and small models
- Explains why deep research favors a smart outer wrapper model
- Hints coding tasks may not benefit from the same fusion setup
Clip transcript
something that it that's in domain or not? Does my question kind of make sense? >> I don't know. Um, I basically I don't it's kind of early to >> um to say. I I think the I wait the results that we published, which are a couple weeks ago, which were focused on deep research, not coding. >> Mhm. >> Um we had the smart model be the wrapper model, be the outer model, and we got the best results from doing that. Um, but it for deep research it works the best. Um, for other tasks, it's like kind of unclear. Like we there it's fusion is not super well optimized for coding. And uh it might be that like a smaller model ends up being like a higher um efficiency per you know, fewer dollars um per like completed successfully completed task, but it's kind of early
Comments
Checking sign-in…
Loading comments…