Moonshot AI Narrows Gap With Top Rivals
The release of Kimi K3.1 Pro places the Chinese model in fourth place on major intelligence benchmarks.
By Malcolm Davis ·
Moonshot AI has introduced Kimi K3.1 Pro, a new reasoning model that has climbed to the fourth position on the Artificial Analysis Intelligence Index. The release marks a significant shift in the competitive landscape for artificial intelligence, as the new system scores 55 points. This places it just three points behind the leading proprietary models, which include Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full.
The most substantial improvements in the new version appear in agentic tasks and coding. On the GDPval-AA benchmark, the model achieves an Elo rating of 1,846, a jump of 312 points over its predecessor. It also records a 65.7 percent score on Terminal-Bench 4.0, a figure that surpasses GPT-6 Astra and trails only slightly behind Claude Opus 5.5. Additionally, Moonshot reports a sharp decline in hallucination rates, dropping from 23.4 percent to 8.3 percent on its internal stress test.
The model demonstrates strong capabilities in 3D modeling and direct interaction with game engines such as Unity and Unreal Engine. It scores 92.7 percent on the BenchCAD benchmark when using external tools, a result that is close to the performance of top competitors. Moonshot notes that the system is particularly effective at parametric modeling for mechanical and architectural designs. Users can now access a feature called 3D Make on the company's website to preview and export these models.
Despite the performance gains, the underlying architecture remains unchanged in size. The model retains 2.8 trillion total parameters with 104 billion active ones and supports a context window of one million tokens. However, the cost of using the system has decreased by 33 percent. Input pricing is now two dollars per million tokens, while output costs ten dollars. Moonshot plans to release the model weights by November 4, 2026, which would make it the highest scoring open weights model available.
The company has also addressed cybersecurity concerns regarding the new release. Moonshot observed that vulnerability research capabilities increased as post-training scaled up. To manage this, the team removed cyber-related data from both training phases for the open weights version. This adjustment reduced the model's success rate on internal vulnerability tests from 90.5 percent to 81.0 percent. The company aims to balance offensive capabilities with defensive utility, noting that open models serve important roles in auditing legacy systems.
Source: The Decoder
