Chinese AI model Kimi K3.1 Pro is closing in on Anthropic's Claude Opus 5.5

A new Chinese AI model, Kimi K3.1 Pro, has been released by Moonshot AI and has nearly matched the performance of Anthropic's Claude Opus 5.5 on the Artificial Analysis Intelligence Index. The model scored 55 points, an 11-point jump over its predecessor, placing it in fourth place on the index. This improvement is significant, as it places the model just three points behind Claude Opus 5.5, the top proprietary model. If Moonshot releases the model weights as planned by November 4, 2026, it would become the highest-scoring open-weights model available, reducing the gap between open and proprietary models from 12 to just 3 points.

The model's performance has been enhanced significantly in agentic tasks, coding, and 3D modeling. On the GDPval-AA benchmark, it reaches 1,846 Elo, which is 312 points above its predecessor, putting it just behind Claude Opus 5.5 and Muse Wonder Full. On Terminal-Bench 4.0, Moonshot reports a score of 65.7 percent, which is close to Claude Opus 5.5's score of 66.4 percent. The model also shows major gains in agentic tasks, with claim-level hallucinations dropping from 23.4 percent with K3 to 8.3 percent with K3.1 Pro.

The parameter count remains at 2.8 trillion, but the API prices have dropped by 33 percent. Token prices are now $2 per million tokens for input, $10 for output, and $0.20 for cache hits, making it more affordable for users. The model is available now through Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.

The company has also launched a feature called '3D Make' on Kimi.com, which allows users to preview, edit, and export 3D models. The model works directly with game engines like Unity, Unreal Engine, and Godot, handling both quick fixes and full project builds. On the BenchCAD benchmark, it scores 92.7 percent with tools, close to Claude Opus 5.5 (96.2 percent) and GPT-6 Astra (95.9 percent).

The model's performance in 3D modeling and coding is impressive, and its affordability makes it a strong contender in the market. The company has also taken steps to address cybersecurity concerns, stripping cyber-related data from both training phases to prevent the model from pushing the absolute frontier of offensive cyber capabilities. Anthropic has shown that even GLM-5.3, which scores 45 on the AA Index, can write exploits at near-Claude Mythos levels. This indicates that the model's performance in cybersecurity is also strong, despite the data cleaning.

The model's performance in cybersecurity is also strong, as it solves 81.0 percent of tasks on its internal xdev-bench-hard test, which is close to DeepSeek-V4.1-Flash's score of 85.7 percent. Kimi K3 scored 33.3 percent on the same test. The company has also taken steps to address cybersecurity concerns, stripping cyber-related data from both training phases to prevent the model from pushing the absolute frontier of offensive cyber capabilities. This indicates that the model's performance in cybersecurity is also strong, despite the data cleaning.