Frontier Models Close 82% of Human Research Gap in Auto-Research Experiment

eliebakouch · x · 2026-08-16

PrimeIntellect released the largest open experiment on auto-research capabilities of frontier models. With 100+ autonomous runs across 10+ models sandboxed on 8xH200s for up to 8 days, the models iterated on the nanoGPT optimizer track. The best runs closed 82% of the performance gap to a record built by human experts over months.

Related event: Prime Intellect's Largest Autonomous AI Research Experiment: 18 Models, 153 Runs, Fable 5 Closes 82% Human Gap(18 posts)→

Original post →

More from Research

• Square lattice connective constant remains unsolved — ripe for AI, argues mathematician — tak3sh8 · 2026-10-16
• Quiz: Double the Residual Stream from 4096 to 8192 — How Does Param Count Change? — Sauers_ · 2026-10-16
• Researcher credits GPT for multiple accepted papers, incl. disproving Wegner's 1965 conjecture — ctjlewis · 2026-10-16
• AutoAgent0: VLM-powered agentic runtime lifts long-tail driving success by 28.3pp — yining_hong · 2026-10-16
• Shuffled labels hit AUC 0.59: selection bias, not signal, drives prosody benchmarks — maier_ak · 2026-10-16
• CODESTRUCT reframes codebases as AST action spaces to fix brittle LLM code agents — burkov · 2026-10-16

Research channel →