Moonshot has released Kimi K3.1 Pro, an upgraded version of its open-weight model, and it is available now in Kimi’s apps and API, the company said.[1][2]

Kimi K3.1 Pro is a post-trained version of Kimi K3, which has 896 individual neural networks with nearly 3 trillion parameters and activates only 16 at a time to reduce inference costs, SiliconANGLE reported.[3] It uses Kimi Delta Attention, a custom implementation of Gated DeltaNet, to reduce the memory used during inference, the company said.[3][2] The upgrade consumed significantly more processing power than prior projects and included automated quality checks and improved infrastructure efficiency, Moonshot said.[3]

On Terminal-Bench 4.0, a test of AI coding agents, K3.1 Pro scored 65.7%, ahead of OpenAI’s GPT-6 Astra at 58.2% and just behind Anthropic’s Claude Opus 5.5 at 66.4%, Moonshot said.[1][2] It scored 92.7% on BenchCAD, a 3D modeling test, approaching Opus 5.5 and Astra, and achieved 64.4 on OfficeQA Pro v2, a financial analysis benchmark, according to Moonshot.[3][2] Moonshot also reported a drop in hallucinations, with the model’s factual error rate falling to 8.3% from 23.4% in Kimi K3 on an internal stress test.[1][2]