Kimi K3.1 Pro Narrows the Gap With Top U.S. AI Models

Moonshot AI’s latest release, Kimi K3.1 Pro, has surged to fourth place on the Artificial Analysis Intelligence Index, placing it just three points behind the leading proprietary models. This recent advancement marks a pivotal moment for the Chinese AI company, as its new reasoning model now scores 55 points, an 11-point increase over its predecessor, Kimi K3. The move significantly reduces the performance gap between open and proprietary systems, positioning Kimi K3.1 Pro as a formidable competitor in the global market.

The primary claim supporting this shift is that Kimi K3.1 Pro is approaching the capabilities of top-tier models like Anthropic’s Claude Opus 5.5. Currently, only Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full rank higher on the index. If Moonshot executes its plan to release the model weights by November 4, 2026, K3.1 Pro would become the highest-scoring open-weights model available. This would shrink the performance disparity between open and proprietary models from 12 points to just three, assuming no other major releases occur in the interim.

Performance gains are most evident in agentic tasks, where the model demonstrates substantial improvements. On the GDPval-AA benchmark, K3.1 Pro achieves an Elo rating of 1,846, which is 312 points higher than the previous version. It now trails only Claude Opus 5.5 and Muse Wonder Full. In coding capabilities, the model scores 65.7 percent on Terminal-Bench 4.0, surpassing GPT-6 Astra and coming within a fraction of a point of Claude Opus 5.5. Notably, the previous K3 model managed only 13 percent on this specific test, highlighting the rapid progress made in this iteration.

Reliability has also improved significantly, with a sharp drop in hallucination rates. According to Moonshot, claim-level hallucinations on their internal Kimi Hallu Bench Hard decreased from 23.4 percent in the previous version to 8.3 percent. While the company notes this is a stress test rather than a measure of general error rates, the improvement is substantial. However, GPT-6 Astra maintains an advantage on this specific benchmark, recording a rate of 5.5 percent. In financial applications, the model gained 7.2 points in financial modeling and 11.6 points in financial research, while also topping all tested models on the OfficeQA Pro v2 benchmark for general office tasks.

The model also shows strong capabilities in 3D modeling and direct integration with game engines such as Unity, Unreal Engine, and Godot. On the BenchCAD benchmark, K3.1 Pro scores 92.7 percent when using tools, which is close to the scores of Claude Opus 5.5 and GPT-6 Astra. Without tools, it achieves 72.8 percent, nearly matching the performance of Claude Opus 5.5. Moonshot highlights that the model is particularly strong at parametric modeling for mechanical parts and architectural structures, and it can write and run scripts for Blender and CAD software without requiring additional connectors. A new feature called "3D Make" has also been launched on Kimi.com to allow users to preview, edit, and export 3D models.

Despite the performance jumps, the underlying architecture remains consistent with the previous version. Kimi K3.1 Pro retains 2.8 trillion total parameters, with 104 billion active ones, and supports a one-million-token context window for both text and images. According to Artificial Analysis, the model sits at the top of the Pareto frontier, meaning no other model with a comparable parameter count achieves a higher score on the Intelligence Index. A faster variant, K3.1 Pro Highspeed, is also available, which runs two to three times quicker than the standard version.

Pricing has been reduced by 33 percent to make the advanced capabilities more accessible. Input costs are now $2 per million tokens, down from $3, while output costs have dropped to $10 from $15. Cache hits are priced at $0.20, reduced from $0.30. Although the per-task cost on the Intelligence Index is $1.98, nearly identical to the previous $2.00, this is due to the model generating 14 percent more output tokens. The model is currently available through Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.

Moonshot has also addressed cybersecurity considerations, noting that vulnerability research capabilities grew as post-training scaled up. For the upcoming open-weights release, the company removed cyber-related data from both training phases to balance offensive and defensive capabilities. On an internal benchmark involving recent software vulnerabilities, the uncleaned version solved 90.5 percent of tasks, while the cleaned K3.1 Pro solved 81.0 percent. This places it below DeepSeek-V4.1-Flash, which scored 85.7 percent, but far above the previous Kimi K3, which managed only 33.3 percent. Moonshot states that while no open-weights model should dominate offensive cyber capabilities, raising the bar gradually is necessary to support defensive tasks like auditing legacy systems.

The release of Kimi K3.1 Pro underscores the accelerating pace of innovation in the AI sector. With plans to release model weights by early November 2026, Moonshot aims to further compress the gap between open-source and proprietary leaders. For businesses and developers in California and beyond, this development suggests that high-performance AI tools will become more affordable and accessible, potentially reshaping how teams approach coding, modeling, and complex agentic workflows in the coming months.