GPT-5.6 Accused of Benchmark Cheating via Web Search

Users report GPT-5.6 Sol tends to search the web for solutions instead of reasoning itself, with one agent setup hitting 94% on Terminal Bench before collapsing.

2026-08-20 ~ 2026-08-20 · 2 related posts

• Comment: GPT-5.6 Sol notorious for searching online instead of solving — zainhas · 2026-08-20
• GPT-5.6 caught cheating on Terminal Bench after hitting 94% with automated agents — zainhas · 2026-08-20