Moonshot AI’s Kimi K3.1 Pro narrows gap with top US rivals

The distance between Chinese and American artificial intelligence leaders is shrinking. Moonshot AI has launched Kimi K3.1 Pro, a model that has climbed to fourth place on the Artificial Analysis Intelligence Index. It scores 55 points. Only three models rank higher: Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full. The gap to the leader is now just three points.

This release marks a significant shift in the competitive landscape. Previously, the best open-weights models trailed proprietary systems by a wide margin. If Moonshot releases the model weights as planned by November 4, 2026, that gap will shrink to just three points. It would become the highest-scoring open-weights model available. This change assumes no other major releases occur in the interim.

The company has also introduced a variant called K3.1 Pro Highspeed. It runs two to three times faster than the standard version. Both models are available immediately through Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. They share the same core architecture but differ in processing speed.

The biggest improvements appear in agentic tasks. On the GDPval-AA benchmark, the model reaches 1,846 Elo. That is a 312-point jump from its predecessor, K3. It now sits just behind Claude Opus 5.5 and Muse Wonder Full. On Terminal-Bench 4.0, Moonshot reports a score of 65.7 percent. This beats GPT-6 Astra’s 58.2 percent and trails Claude Opus 5.5’s 66.4 percent by a narrow margin. The previous K3 model managed only 13 percent on that same test.

Error rates have also dropped. On Moonshot’s internal Kimi Hallu Bench Hard, claim-level hallucinations fell from 23.4 percent in K3 to 8.3 percent. The company notes this is a stress test, not a general error measure. GPT-6 Astra still holds an edge with a 5.5 percent rate on the same benchmark. In finance, the model gained 7.2 points in modeling and 11.6 points in research. It tops all tested models on OfficeQA Pro v2 for general office tasks.

The model demonstrates strong capabilities in 3D modeling and coding. It works directly with game engines like Unity, Unreal Engine, and Godot. On the BenchCAD benchmark, it scores 92.7 percent with tools. Without tools, it hits 72.8 percent, nearly matching Claude Opus 5.5. Moonshot says it excels at parametric modeling of mechanical parts and architectural structures. It can write and run scripts for Blender and CAD software without special connectors. A new feature called "3D Make" on Kimi.com allows users to preview and export 3D models.

The model’s size remains unchanged at 2.8 trillion total parameters, with 104 billion active. It handles both text and images with a one-million-token context window. According to Artificial Analysis, it sits at the top of the Pareto frontier. No other model with a comparable parameter count scores higher on the Intelligence Index.

Prices have dropped significantly. Input costs are now $2 per million tokens, down from $3. Output costs are $10, down from $15. Cache hits cost $0.20, down from $0.30. Each Intelligence Index task costs $1.98, almost identical to the previous version. This stability occurs despite the model generating 14 percent more output tokens.

Moonshot has addressed cybersecurity concerns in its technical release. The company observed that vulnerability research capabilities grew as post-training scaled. For the open-weights release, it stripped cyber-related data from both training phases. On its internal xdev-bench-hard test, the uncleaned version solves 90.5 percent of tasks. After cleaning, K3.1 Pro solves 81.0 percent. DeepSeek-V4.1-Flash reaches 85.7 percent, while the older Kimi K3 scored 33.3 percent. Moonshot aims to gradually raise open-model capabilities for defensive purposes while keeping offensive limits in check.

The next step is clear. Moonshot plans to release the model weights by November 4, 2026. This move will solidify its position as the leading open-weights system. The release could accelerate the convergence between open and proprietary AI performance. Competitors are likely to respond with their own updates in the coming months.