🚀 Kimi K3.1 Pro is here, and the benchmarks are impressive! Moonshot AI's new 3T-class open-weight model comes with a 1 MILLION token context window and major improvements in coding, 3D creation, and agentic capabilities. Some highlights: 🔥 65.7% on Terminal-Bench 4.0, beating GPT-6 Astra (58.2%) and nearly matching Claude Opus 5.5 (66.4%). 🔥 Strong 3D/CAD capabilities, including generating complex models directly from sketches and text. 🔥 Major improvements in financial research and modeling, with substantially reduced hallucination rates. 🔥 New Highspeed variant delivering 2–3x faster inference. K3.1 Pro is available now through Kimi, Kimi Code, Kimi Work, and the API. Full model weights will follow on November 4. Another very serious contender from China. The open-weight frontier keeps advancing!