According to ChainCatcher citing Phoenix Net Tech, Moonshot AI announced on November 5 that it had released the Kimi K3.1 Pro model weights on Hugging Face. Arena ranked Kimi K3.1 Pro (Max) first among open-weight models and 13th overall on its Agent Arena leaderboard, with top open-model rankings in work, coding, and chat. The rankings drew on more than 17,500 real-world, long-horizon agent-task sessions. The model posted a net improvement rate of +5.1%, compared with +0.9% for the previous Kimi K3 (Max); its Bash Reliability score was +7.9%, versus -4.5% for its predecessor. Moonshot AI says the model uses a mixture-of-experts architecture with 2.8 trillion total parameters and about 104 billion active parameters per inference, supports a context window of about 1 million tokens, and has native multimodal capabilities.