This YouTube interview clip (May 22, 2024) discusses how to choose LLMs for AI agents. The speakers emphasize trade-offs among reliability, speed, and cost, note that fine-tuning and other techniques can improve agents, and recommend using the most powerful (even slow) model for early-stage ideation and new design patterns, then optimizing or open-sourcing the approach later.
Is there a model that you think would work best as an agent, or are they all kind of similar? What are your thoughts there?
Picking the right model is extremely helpful in getting your agent to be more reliable, faster, cheaper. And then of course, there's fine tuning, and there's a lot of techniques you can do. For me personally, I'm much more interested in ideating around new design patterns, and for that it's easiest to just use the most powerful model, even if it's slow, just to like get it working. And then if that new design pattern works, then I in theory could go back and optimize the model model the prompt. But oftentimes instead of doing that, I just open source it and move on to the next design pattern.
AI News: Dots, GPT-6.1 Sol, Sonnet 5.5, Gemini 4, and everything you need to know
