We're proud to announce that Anthropic's latest Sonnet 4.5 model has now been incorporated into Capsa AI.

In what is becoming a battle of margins, Anthropic claims the top spot again across benchmark evaluations. Across STEM, Law, Medicine, and Finance evaluations, Finance use cases show the biggest improvement versus their last release (just 2 months ago).

Our initial benchmarks confirm Sonnet 4.5 as state-of-the-art versus the competition, and it's now core to our platform (and also the model of choice for our team daily).

Why This Matters

Less than 2 months after GPT-5's release, the race for AI improvements is getting increasingly marginal. To the untrained eye, there's little difference in day-to-day outputs. However, in autonomous detailed workflows, the step-change is exceedingly clear: agents follow multi-step analysis more diligently, with fewer hallucinations and faster responses.

At Capsa, we strongly believe in our multi-model approach. Through thorough research and engineering from our team, our users never have to worry about accessing the latest models.