upvote

points

by rohaga5 hours ago |
comments
upvoteby brandall103 hours ago|
next
[-]
Seems like it might be more advantageous to just adjust reasoning effort to retain cache. Maybe in some exceptional cases where there will be a ton more inference to solve the problem, but going significantly dumber in that case seems counterintuitive.

I can really only see the utility of things like spawning subagents to a lower tier model from another provider, and that's something harnesses can already handle (ie. give model specs for certain delegation roles).

reply
upvoteby rohaga2 hours ago|
parent|
next
[-]
I agree that adjusting the reasoning effort to retain cache is a huge thing! But even doing that automatically is currently a challenge for people to figure out and do well, and costs mental energy when perhaps it doesn't need to. For example, there is GPT-5.6-Sol low, med, high, xhigh, max, and lots of "rules of thumb" that people develop on which one to use when.
reply
upvoteby verdverm3 hours ago|
parent|
prev|
[-]
caching is per model, it does not transfer between them
reply
upvoteby brandall103 hours ago|
parent|
[-]
Updated to clarify reasoning effort.
reply
upvoteby ignoramous4 hours ago|
prev|
[-]
> the next task is SO HARD that it doesn't make sense to not use frontier

How does tokenless quantify "so hard"?

> 1 turn to crack versus 100 turns for Deepseek to crack

Interesting definition for a "frontier". What is a "turn" here? Token count? Request count? Context-based?

I've found that MiniMax M3 (a smaller model at 295b) will code up better when DeepSeek v4 Pro (1.6t) will not (and vice versa).

reply
upvoteby rohaga4 hours ago|
parent|
[-]
Highly encourage you to read the blog post (https://usetokenless.com/blog/building-tokenless). Essentially, we estimate the confidence of a specific model failing or succeeding on a specific task using our own foundation models.

A turn here is a tool call/user input, anything that causes the model to get some new input. We're working on adding Minimax M3 and other models. We think that people have some intuitions about which models are good when--we seek to quantify them scientifically.

reply