Moonshot AI has launched an upgraded Kimi model, Kimi K3.1 Pro, touting improved coding, financial research and 3D design capabilities while deliberately limiting the model’s ability to carry out cyberattacks.
Kimi K3.1 Pro debuted fourth on independent evaluator Artificial Analysis’ Intelligence Index, scoring 55 points—an 11-point increase over its predecessor—and, the evaluator said, would become the highest-ranked open-weight model on the index once its weights are published. The first model in the K3.1 series is available through Kimi apps, coding tools and the company’s developer platforms; Moonshot plans to release the full model weights by Nov. 4.
Kimi K3.1 Pro performance and technical details
Kimi K3.1 Pro has 2.8 trillion parameters, with roughly 104 billion parameters activated for each token processed, and a one-million-token context window that enables the handling of large volumes of information in a single session. Moonshot said the upgrade was built through further training of Kimi K3, with improvements aimed at completing practical tasks—from building software and creating 3D assets to analysing financial documents and executing extended, multistep assignments.
To limit cyber risk, Moonshot removed cybersecurity-related data from parts of its training pipeline, including reinforcement-learning environments, while retaining general programming tasks. “We aim not to release open weights that would extend the absolute frontier of cyber capability,” Moonshot said.
On an internal test of vulnerability discovery and exploitation, the company reported K3.1 Pro scored 81.0 percent, compared with 90.5 percent for a comparable version without such safeguards; those figures were self-reported by Moonshot.
In coding performance measured by Terminal-Bench 4.0, K3.1 Pro scored 65.7 percent—ahead of OpenAI’s GPT-6 Astra at 58.2 percent and just behind Anthropic’s Claude Opus 5.5 at 66.4 percent, according to the company’s evaluation.
Bloomberg reported that Moonshot, a Beijing-based firm, recently completed a private funding round valuing it at about $50 billion and is targeting a Hong Kong debut in the first quarter of 2027, citing people familiar with the matter.
China’s models closing gap with US rivals
A Bloomberg Intelligence report said China’s top AI models have narrowed the performance gap with US rivals to around 3 percent. Bloomberg Intelligence senior analyst Robert Lea said the speed of the catching-up process “raises fresh questions over how long the US can maintain its technological lead, and could help Chinese AI firms capture a larger share of the global market.”
Stanford’s 2026 AI Index said the performance gap between the top US and Chinese models had “effectively closed”: as of March the best US model remained ahead by 2.7 percent, and models from the two countries had traded places at the top of rankings several times since early 2025.
Tsinghua University figures reflected a mixed assessment of China’s position. Zhang Yaqin, founding dean of Tsinghua’s Institute for AI Industry Research, said Chinese models have entered the global top tier and China is carving out its own path in areas such as model efficiency, open-source development and industrial applications. Tang Jie, a Tsinghua computer science professor, said China still needs to catch up in high-end computing chips, fundamental AI architecture and the broader technology ecosystem, and must produce more breakthroughs that originate in China rather than primarily improving on existing approaches.
