Moonshot AI Releases Kimi K3.1 Pro to Challenge Top Global Models

Moonshot AI has officially launched Kimi K3.1 Pro, a new reasoning model that immediately secures the fourth position on the Artificial Analysis Intelligence Index. The release marks a significant leap for the Chinese company, which has positioned this new system as a direct competitor to the leading proprietary models currently dominating the market. The move signals an intensifying race among AI developers to close the performance divide between open and closed systems.

The performance metrics for K3.1 Pro are substantial. The model achieved a score of 55 on the Intelligence Index, representing an 11 point increase over its predecessor, Kimi K3. This places it only three points behind the top tier of proprietary systems. Currently, only Anthropic's Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full outperform it, with scores of 58, 56, and 56 respectively. If Moonshot follows through on its plan to publish the model weights by November 4, 2026, K3.1 Pro would become the highest scoring open weights model available. That release would reduce the gap between open and proprietary models from 12 points to just three, assuming no other changes occur in the meantime.

The company is offering two primary versions of the model to address different user needs. Alongside the standard K3.1 Pro, Moonshot is providing a Highspeed variant that the company states runs two to three times faster. Both versions are accessible through the company's existing platforms, including Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. This wide availability ensures that developers and businesses can integrate the new model into their workflows immediately without waiting for a single release channel.

The most dramatic improvements in K3.1 Pro are found in agentic tasks. On the GDPval-AA benchmark, the model reached an Elo rating of 1,846, which is 312 points higher than K3. It trails only Claude Opus 5.5 and Muse Wonder Full in this specific category. In terminal environments, the model reported a score of 65.7 percent on Terminal-Bench 4.0. This result beats GPT-6 Astra, which scored 58.2 percent, and sits slightly below Claude Opus 5.5 at 66.4 percent. By comparison, the older K3 model managed only 13 percent on that same test.

Accuracy has also seen a major upgrade, with a sharp reduction in hallucinations. On the company's internal Kimi Hallu Bench Hard, claim level hallucinations dropped from 23.4 percent with K3 to just 8.3 percent. Moonshot describes this benchmark as a stress test rather than a general error rate measure. It notes that GPT-6 Astra still leads this metric with a 5.5 percent rate. In financial applications, the model gained 7.2 points on Financial Modeling and 11.6 points on Financial Research. The company also claims that K3.1 Pro tops all tested models on the OfficeQA Pro v2 benchmark.

Capabilities in 3D modeling and coding have expanded significantly. K3.1 Pro can now work with game engines such as Unity, Unreal Engine, and Godot. It can write and run scripts for Blender and CAD software without the need for MCP connectors. On the BenchCAD benchmark, the model scores 92.7 percent when using tools. This compares to 96.2 percent for Claude Opus 5.5 and 95.9 percent for GPT-6 Astra. Without tools, the model reaches 72.8 percent, nearly matching the 73.0 percent score of Claude Opus 5.5. A new 3D Make feature on Kimi.com now allows users to preview, edit, and export 3D models directly.

The underlying architecture remains consistent with K3, featuring 2.8 trillion total parameters with 104 billion active. The model supports a one million token context window and handles both text and images. Artificial Analysis places it at the top of the Pareto frontier, indicating no model of comparable size scores higher on their index. Pricing has been reduced by 33 percent, with input costs at $2 per million tokens, output at $10, and cache hits at $0.20. Although the model generates 14 percent more output tokens, the cost for each Intelligence Index task remains nearly identical to K3 at $1.98 compared to $2.00.

Moonshot has addressed cybersecurity concerns in its technical documentation. The company explains that vulnerability research capabilities grew because cyber related content was part of the general mid training corpus and included in the reinforcement learning mix. For the open weights release, the team removed this data from both training phases. On the internal xdev-bench-hard, which reproduces recently disclosed vulnerabilities in software like the Linux kernel and Node.js, the uncleaned model solved 90.5 percent of tasks. After cleaning, that figure dropped to 81.0 percent. DeepSeek-V4.1-Flash scored 85.7 percent, while Kimi K3 managed only 33.3 percent. Moonshot aims to gradually raise the capabilities of open models for both offensive and defensive work, noting that even lower scoring models can write exploits at levels close to top proprietary systems.