Qwen4-Max jumps 27.9 to 46.7 on Agents' Last Exam, topping Claude and GPT

AccBalanced · x · 2026-11-16

A benchmark chart shared by Canopy Wave AI shows Qwen4-Max leaping from 27.9 to 46.7 on Agents' Last Exam v1.1 (long-horizon tasks), overtaking Claude (43.4), GPT (42.6), and other frontier models to take the top spot — another strong third-party result following the model's release.

Original post →

More from Models

Models channel →