Moonshot AI released Kimi K3.1 Pro on October 23, 2026, a 2.8T-total/104B-active MoE model with 1M-token context that immediately ranked fourth on the Artificial Analysis Intelligence Index at 55 points. It posted strong self-reported results on agentic and coding benchmarks such as Terminal-Bench 4.0 (65.7%) and HLE variants with tools (70.0%), placing it just behind Claude Opus 5.5 while leading some proprietary models on tasks like SWE-Marathon and OfficeQA Pro v2. The company deliberately limited cyber capabilities in the forthcoming open weights, scheduled for November 4, which could enable independent verification on HLE-Diamond’s 1,000-question subset. Traders are watching whether the model’s post-training gains translate to a competitive or leading position on that refined leaderboard amid rapid frontier-model releases.