Moonshot AI's Kimi K3.1 Pro narrows the gap with leading US models

Moonshot AI's new Kimi K3.1 Pro model has secured the fourth spot on the Artificial Analysis Intelligence Index with a score of 55 points. This places it just three points behind Anthropic's Claude Opus 5.5. The update represents a major shift in the competitive landscape for large language models.

The company released K3.1 Pro as a post-trained version of its predecessor. It utilized significantly more compute during the post-training phase. This approach drove an 11-point increase over the original K3 model. Only three models currently rank higher. These are Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full. If Moonshot releases the model weights by November 4, 2026, K3.1 Pro would become the highest-scoring open-weights model available. Such a move would shrink the performance gap between open and proprietary systems from 12 points to just 3.

Performance improvements are most evident in agentic tasks. On the GDPval-AA benchmark, the new model reaches an Elo score of 1,846. This is a 312-point increase over K3. It trails only Claude Opus 5.5 and Muse Wonder Full in this category. In coding tests, Moonshot reports a 65.7 percent score on Terminal-Bench 4.0. This result surpasses GPT-6 Astra and comes within a fraction of Claude Opus 5.5. The previous K3 model managed only 13 percent on that same test.

The model also shows better reliability in specific domains. Hallucination rates dropped sharply on internal benchmarks. Claim-level errors fell from 23.4 percent to 8.3 percent. Moonshot notes this is a stress test rather than a general error measure. GPT-6 Astra still holds an edge with a 5.5 percent error rate. The model also handles 3D modeling and coding for game engines like Unity and Unreal. It scores 92.7 percent on the BenchCAD benchmark with tools, closely matching its competitors.

Pricing and structural dimensions remain a key differentiator. The model retains the same 2.8 trillion total parameters and 104 billion active parameters as K3. The context window spans one million tokens. Despite this size, API prices drop by 33 percent. Input costs are now $2 per million tokens, while output costs stand at $10. This reduction aims to make the high-performance model more accessible for developers and businesses.

Moonshot addressed cybersecurity capabilities in its technical release. The company observed that vulnerability research abilities grew as post-training scaled. For the open-weights release, they removed cyber-related data from the training process. This adjustment lowered the model's success rate on internal security benchmarks from 90.5 percent to 81.0 percent. The company believes this balance allows open models to serve defensive purposes without dominating the offensive frontier.

A Highspeed variant of the model is also available, running two to three times faster. Both versions are accessible through Kimi.ai, Kimi Work, Kimi Code, and the API. The release signals a continued push by Chinese developers to close the performance gap with top US proprietary models while maintaining cost efficiency.