Moonshot AI’s July 2026 launch of the 2.8-trillion-parameter Kimi K3 model has anchored trader sentiment for the Polymarket on peak Kimi HLE performance this year. K3 has posted scores as high as 56% with tools on public leaderboards and around 43.5–47% without, placing it just behind leading Claude and GPT-5.5 variants on the challenging 2,500-question benchmark of expert-level academic reasoning. Chinese labs continue rapid iteration on large open-weight systems that close gaps on frontier closed models, while no newer Kimi releases have surfaced since the July debut. Traders are watching for potential year-end updates or training runs that could push scores toward or past the 55–60% thresholds still dominated by Anthropic and OpenAI systems.