Kimi K3.1 Pro Launches at 4th on Artificial Analysis Intelligence Index With Score 55

Moonshot released Kimi K3.1 Pro, a reasoning model scoring 55 on the Artificial Analysis Intelligence Index, placing fourth and gaining 11 points over Kimi K3. The model posts large agentic gains, including 1,846 Elo on GDPval-AA and 67% on Terminal-Bench 4.0, and is set to become the highest-scoring open-weights model once its weights are published. It carries a 1M-token context window, 2.8T total and 104B active parameters, and pricing of $2/$10/$0.20 per 1M input/output/cache-hit tokens.
Event version 1.
Key insights
Fourth on Intelligence Index
Kimi K3.1 Pro scores 55 on the Artificial Analysis Intelligence Index, up 11 points from Kimi K3 at 44. It trails Claude Opus 5.5 at 58, Claude Sonnet 5.5 at 56, and Muse Wonder Full at 56.
Agentic performance gains
Kimi K3.1 Pro scores 1,846 Elo on GDPval-AA, up 312 Elo from Kimi K3, placing third behind Claude Opus 5.5 and Muse Wonder Full. It achieves the highest score of any model on Terminal-Bench 4.0 at 67%, up from 13% for Kimi K3, and AutomationBench-AA rises from 58% to 73%.
Moderate gains on other evals
Humanity's Last Exam rises from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. The AA-Omniscience Index rises from 20 to 30 as the hallucination rate falls from 53% to 44% and accuracy rises from 48% to 51%.
Pricing and cost per task
Cost per task is roughly equal to Kimi K3 at $1.98, with K3.1 Pro using 55K output tokens per Intelligence Index task versus 48K for Kimi K3. First-party token prices are 33% lower at $2/$10/$0.20 per 1M input/output/cache-hit tokens.
Open-weights gap to narrow
Once K3.1 Pro weights are released, it will be the highest-scoring open-weights model and the gap between leading open-weights and proprietary models on the Intelligence Index will narrow from 12 points to 3. Weights are currently not published.