Moonshot AI Officially Releases Kimi K3.1 Pro 3-Trillion Parameter Model
Moonshot AI has officially launched Kimi K3.1 Pro, its most powerful 3-trillion parameter model to date, along with a high-speed variant featuring a 1-million context window. The model is immediately accessible across multiple platforms and APIs, with full model weights scheduled for release by November 4. Kimi K3.1 Pro achieves or approaches the performance levels of top-tier closed-source models in coding, agents, and complex reasoning benchmarks. This release highlights the rapid advancements of domestic AI companies in challenging global frontier models across specialized technical workflows. The model achieved 65.7% on Terminal-Bench 4.0, outperforming GPT-6 Astra, and demonstrated robust capabilities in 3D modeling, parametric design, and end-to-end financial workflows. Furthermore, its factual error rate in stress tests dropped significantly compared to its predecessor K3.
## BACKGROUND
Terminal-Bench is an open-source evaluation benchmark designed to measure the performance of AI agents on complex, real-world tasks within command-line and terminal environments. Benchmarks like this help developers rigorously test how well autonomous agents handle multi-step tool usage and software engineering challenges.