Feature Idea
When building a new agent, selecting the best model is not easy. I tend to use GPT4o by default for all assistants because it’s time-consuming (and hard) to compare all models. I found this article on Dust blog which compares the models, and I’m wondering if it would be feasible to get an automated recommendation of the best model when building an agent, based on its description. I understand it’s not possible to know for sure which one is the best, yet there might still be valuable recommendations based on the available knowledge about models performances.
For example, I’m about to build an assistant which can write and run SQL queries on our database, and I’m pretty sure there must be existing knowledge about which models have the best performance for this. Having the “auto” mode in Dust would feel great !
