Artificial analysis put the Claude Fable 5.2 through its benchmark: 65 points and first place, 5 above the Gemini 4 Carbon and 6 above the Opus 5.5.
The model is the strongest in science and knowledge. There is no gap in the performance of the agents: in Terminal-Bench 4.0 it has the same 60% as Opus 5.5, and Kimi K3.1 Pro, Gemini 4 Carbon and Sonata 5.5 are ahead.
Max Mission costs $26.5, 2.6 times more expensive than opus 5.5. High is more profitable: 62 points, more than any other model, and about $9 per mission. Colleagues took a closer look.