If you are training on data labeled by frontier models, how do you expect to exceed the performance of frontier models, other than in the cost dimension by recognizing simpler problems and routing to cheaper models?
If you are training on data labeled by frontier models, how do you expect to exceed the performance of frontier models, other than in the cost dimension by recognizing simpler problems and routing to cheaper models?
Different frontier models are good at different things! We'll be the ones combining them optimally.
I understand the premise, but I believe success depends on you being able to effectively sort problems for other frontier models better than the frontier.
Yes! We believe this is entirely possible. Frontier LLMs are trained to solve different problems. We’re training a frontier router :)