TEMPO System Achieves Perfect IMO 2026 Score with Self-Critique Reasoning

Scobleizer · x · 2026-08-17

A new system called TEMPO achieved a perfect 42/42 score on the IMO 2026. It utilizes a unique architecture where a single model acts as both actor and critic, pausing mid-thought to evaluate progress. Unlike simply expanding context windows, TEMPO actively rips apart its own proofs until they hold. The model features 280B total parameters (16B active), 512K context, and multimodal capabilities.

Original post →

More from AGI Musings

AGI Musings channel →