François Chollet Quit Google to Build the AI Test LLMs Can’t Pass — Humans 100%, GPT-5.5 0.43%By Chew Loong Nian - AI ENGINEER / May 3, 2026 GPT-5.5 — the model OpenAI shipped two weeks ago, the one rated 1501 Elo on LMArena — scored 0.43%.Continue reading on Towards AI »