55 points: Moonshot's Kimi K3.1 Pro narrows the gap with top US AI models

Moonshot AI has released Kimi K3.1 Pro, a model that now sits in fourth place on the Artificial Analysis Intelligence Index with a score of 55. This places it within striking distance of the current market leaders, specifically Anthropic's Claude Opus 5.5, which holds the top spot with 58 points. The new release marks a significant step forward for the Chinese developer, which has been working to close the performance gap with major American competitors.

The financial implications of the launch are immediate for developers and enterprises. While the model retains the same 2.8 trillion parameter architecture as its predecessor, Moonshot has reduced API pricing by 33 percent. Input costs have fallen to $2 per million tokens, down from $3, while output costs are now $10 per million, reduced from $15. Cache hits are priced at $0.20. Despite the lower price point, the cost per Intelligence Index task remains nearly identical at $1.98, primarily because the new model generates 14 percent more output tokens to complete tasks.

Performance gains are most visible in complex, agentic workflows. On the GDPval-AA benchmark, Kimi K3.1 Pro achieved an Elo rating of 1,846, a 312 point improvement over the previous version. It trails only Claude Opus 5.5 and Muse Wonder Full in this specific metric. Coding capabilities have also seen a dramatic leap. The model scored 65.7 percent on Terminal-Bench 4.0, surpassing GPT-6 Astra and coming within less than one point of Claude Opus 5.5. This is a stark contrast to the previous K3 model, which managed only 13 percent on the same test.

The company also highlighted improvements in reliability and specialized tasks. Hallucination rates on internal stress tests dropped from 23.4 percent to 8.3 percent, though GPT-6 Astra maintains a lower rate of 5.5 percent. In the 3D modeling sector, Kimi K3.1 Pro integrates directly with engines like Unity and Unreal Engine. It scored 92.7 percent on the BenchCAD benchmark, closely matching the performance of top US rivals. Moonshot has also introduced a new feature called 3D Make, which allows users to preview and export models directly through the Kimi platform.

Strategic decisions regarding open weights will play a crucial role in the coming weeks. Moonshot plans to release the model weights by November 4, 2026. If this schedule holds, Kimi K3.1 Pro would become the highest scoring open weights model available. This would shrink the performance gap between open and proprietary systems from 12 points to just three. The company has also addressed cybersecurity concerns, noting that it stripped cyber related data from the training set for the open release to balance offensive and defensive capabilities.

For New England businesses looking to integrate advanced AI, the entry now offers a compelling balance of cost and capability. With a one million token context window and support for both text and images, the model is positioned at the top of the Pareto frontier for its size. The release underscores a broader trend in the AI market where non US players are increasingly competitive on both price and performance, forcing established leaders to continue rapid innovation to maintain their edge.