Tencent Hunyuan Hy4 tops APEX-Agents open-weights leaderboard with 77.6% Pass@1
ShunyuYao12 · x · 2026-11-06
Mercor's APEX-Agents leaderboard has a new open-weights #1: Tencent Hunyuan's Hy4 at 77.6% Pass@1, ahead of DeepSeek-V4.1-Flash-1017 (69.8%), Kimi K3.1 Pro (65.5%), Qwen3.8-Max (63.3%) and MiMo V2.6 Flash RL (57.4%). Hy4 leads every open-weights domain: Management Consultant 85.6%, Investment Banking Analyst 75.3%, Corporate Lawyer 71.9%. The APEX family tests agents on economically valuable professional work, including a Harvey-partnered legal benchmark of 1,749 real attorney-built tasks.
More from Models
- On the METR graph: insiders seeing AECI tasks is weak evidence, not zero evidence — gleech · 2026-11-06
- Daily AI brief: GPT-6.2 imminent, Claude Personal Finance beta, METR says Anthropic progress up 1.9x — testingcatalog · 2026-11-06
- METR report researcher quotes lean bullish: Opus 4.5 coding plus faster experiment cycles — gleech · 2026-11-06
- gleech: dropping the three weak benchmarks raises the estimated slope, but the y-axis is meaningless — gleech · 2026-11-06
- Epoch AI review: OSWorld tasks are simple, unstable, and barely test GUI use — gleech · 2026-11-06
- Steering a single MoE expert makes Qwen3 claim it can "see the screen" — Sauers_ · 2026-11-06