Moonshot releases Kimi K3.1 Pro and says it cut its hacking skills

Moonshot AI has released Kimi K3.1 Pro, an upgraded version of Kimi K3. The model nearly matches Claude Opus 5.5 in coding performance, scoring 65.7% on Terminal Bench 4.0 compared to 66.4%. Full weights are expected by November 4. To prevent misuse of the open release, Moonshot removed hacking-related training data from the model. This update positions the Chinese AI lab's latest model as a strong competitor in technical coding benchmarks while prioritizing safety.