Kimi K3: second only to Fable 5 on AA-Briefcase
Summary
Kimi K3 from Moonshot AI is a 2.8T parameter model that places second on the AA-Briefcase agentic knowledge-work benchmark, with an Elo of 1543 behind Claude Fable 5. The article highlights strong analytical performance but weaker presentation, along with a high cost per task and long average time per task, illustrating trade-offs in frontier AI models for enterprise use.