Kimi K3.1 Pro Ranks Fourth on Intelligence Index

Moonshot launched Kimi K3.1 Pro on October 23, 2026, with a score of 55 on the Artificial Analysis Intelligence Index, placing it fourth overall. The reasoning model gained 11 points over Kimi K3 (max, 44), behind Claude Opus 5.5 (max with fallback, 58), Claude Sonnet 5.5 (max with fallback, 56), and Muse Wonder Full (max, 56). Its weights have not been published; Artificial Analysis says it will lead open-weights models once Moonshot releases them.

Kimi K3.1 Pro scored 1,846 Elo on GDPval-AA, third overall and 312 Elo above Kimi K3, behind Claude Opus 5.5 at 1,866 and Muse Wonder Full at 1,857. It scored 67% on Terminal-Bench 4.0, the highest result of any model, up from Kimi K3's 13% and ahead of Claude Sonnet 5.5's 64%. On AutomationBench-AA, which tests agentic SaaS workflows, its score rose from 58% to 73%.

Results also rose on other evaluations: Humanity's Last Exam from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. The AA-Omniscience Index increased from 20 to 30 as the hallucination rate fell from 53% to 44% and accuracy rose from 48% to 51%. Artificial Analysis estimates its cost at $1.98 per task, roughly equal to Kimi K3. It uses 55K output tokens per Intelligence Index task, 14% more than Kimi K3's 48K, while first-party token prices are 33% lower.

The model has a 1M-token context window, 2.8T total parameters and 104B active parameters, and accepts text and images. Its listed prices are $2/$10/$0.20 per 1M input/output/cache-hit tokens, 33% below Kimi K3 on all three rates. Artificial Analysis says that after the weights are released, the gap between the leading open-weights and proprietary models on the Intelligence Index will narrow from 12 points to 3.

Moonshot launched Kimi K3.1 Pro on October 23, 2026, with a score of 55 on the Artificial Analysis Intelligence Index, placing it fourth overall. The reasoning model gained 11 points over Kimi K3 (max, 44), behind Claude Opus 5.5 (max with fallback, 58), Claude Sonnet 5.5 (max with fallback, 56), and Muse Wonder Full (max, 56). Its weights have not been published; Artificial Analysis says it will lead open-weights models once Moonshot releases them.

Kimi K3.1 Pro scored 1,846 Elo on GDPval-AA, third overall and 312 Elo above Kimi K3, behind Claude Opus 5.5 at 1,866 and Muse Wonder Full at 1,857. It scored 67% on Terminal-Bench 4.0, the highest result of any model, up from Kimi K3's 13% and ahead of Claude Sonnet 5.5's 64%. On AutomationBench-AA, which tests agentic SaaS workflows, its score rose from 58% to 73%.

Results also rose on other evaluations: Humanity's Last Exam from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. The AA-Omniscience Index increased from 20 to 30 as the hallucination rate fell from 53% to 44% and accuracy rose from 48% to 51%. Artificial Analysis estimates its cost at $1.98 per task, roughly equal to Kimi K3. It uses 55K output tokens per Intelligence Index task, 14% more than Kimi K3's 48K, while first-party token prices are 33% lower.

The model has a 1M-token context window, 2.8T total parameters and 104B active parameters, and accepts text and images. Its listed prices are $2/$10/$0.20 per 1M input/output/cache-hit tokens, 33% below Kimi K3 on all three rates. Artificial Analysis says that after the weights are released, the gap between the leading open-weights and proprietary models on the Intelligence Index will narrow from 12 points to 3.