FrontierMath Tier 4 saturates in Sept 2026, a year ahead of Epoch AI's forecasts
ben_j_todd · x · 2026-09-11
Ben Todd notes Epoch AI's math benchmark forecasts were more accurate than almost everyone's, yet still over a year too slow: FrontierMath Tier 4 saturated in September 2026, about a year earlier than Epoch expected for the easier Tier 1-3. AI is already solving substantive lemmas.
The linked Epoch report (commissioned by Google DeepMind) argues:
- Scaling likely persists to 2030, requiring hundreds of billions in investment and gigawatts of power
- Lab revenue growth would justify hundred-billion-dollar scaling investments
- Extrapolation suggests by 2030 AI will implement complex scientific software from natural language, assist proof formalization, and answer open-ended biology questions
Related event: Ben Todd's Forecast Review: Everyone Is Underestimating AI Progress(6 posts)→
More from AGI Musings
- Anthropic's S-1 pledges AI for curing diseases and serving 4.6B underserved patients — petergyang · 2026-10-27
- Goldman finds statistically significant negative link between AI adoption and US employment growth — ben_j_todd · 2026-10-27
- a16z: World-changing tech makes a chaotic entrance — AI agents are following the same path — a16z · 2026-10-27
- NEJM AI on medical education: mastery requires performing without the machine — BraydonDymm · 2026-10-27
- Dev warns abliterated models could wreck software supply chains with 1000+ transitive dependencies — nitarshan · 2026-10-27
- Delete your priors: why every SOTA model drop should reset how you think the world works — zack_overflow · 2026-10-27