We’ve recently updated the models for SquiggleAI, adding Sonnet 4.5, Haiku 4.5, and Grok Code Fast 1.Initial tests show promising results, though probably not game-changing. Will run further tests on numeric differences.People are welcome to use (for free!) It’s a big unreliable, so feel free to run a few times on each input.
https://quantifieduncertainty.org/posts/updated-llm-models-for-squiggleai/
We’ve recently updated the models for SquiggleAI, adding Sonnet 4.5, Haiku 4.5, and Grok Code Fast 1.
Initial tests show promising results, though probably not game-changing. Will run further tests on numeric differences.
People are welcome to use (for free!) It’s a big unreliable, so feel free to run a few times on each input.
https://quantifieduncertainty.org/posts/updated-llm-models-for-squiggleai/