Moonshot released Kimi K3.1 Pro on 23 October. Artificial Analysis scored it 55 on its Intelligence Index, placing it fourth overall and eleven points above K3's 44. API pricing is $2 per million input tokens and $10 per million output, with cache hits at $0.20 - all three rates a third below K3's $3 and $15. The model is a mixture-of-experts with 2.8 trillion total parameters, 104 billion active, and a 1 million token context window.
It sits behind Claude Opus 5.5 at 58, with Claude Sonnet 5.5 and Muse Wonder Full tied at 56. On GDPval-AA it scored 1,846 Elo against Opus 5.5's 1,866 and Muse Wonder Full's 1,857. The two Terminal-Bench 4.0 figures in circulation are both real and measure different things: Artificial Analysis independently recorded 67%, ahead of Sonnet 5.5's 64%, while Moonshot self-reported 65.7% against Opus 5.5's 66.4%. Moonshot's own sheet also claims 51.7 on SWE-Marathon against Opus 5.5's 46.3, and 64.4 on OfficeQA Pro v2. It is well behind on science: 38.1 on Terminal-Bench Science 0.1 against GPT-6 Astra's 68.1.
The more unusual detail is what Moonshot removed. For the open-weight release it stripped cyber-related data from mid-training and reinforcement learning. On its internal xdev-bench-hard the shipped model solved 81.0% at pass@3, down from 90.5% for an unfiltered checkpoint - a deliberate 9.5-point cut to its own model's offensive capability before publishing the weights. Moonshot also reports claim-level hallucinations falling from 23.4% on K3 to 8.3% on its internal hard hallucination benchmark. Weights were not published at launch; Moonshot says November, at which point this becomes the highest-scoring open-weights model available.
Sources
Artificial Analysis · MarkTechPost · TNW