Replication of Epoch benchmark index says AI could beat the human baseline by October 2026
herbiebradley · x · 2026-08-04
A replication of Epoch’s capabilities index suggests that the human baseline across 44 benchmarks could be overtaken by AI models around October 2026.
- The personal replication, called BECI, puts the human score at 166.7 across 44 benchmarks with a human baseline.
- It projects models crossing that line in about two months (roughly October 2026).
- For comparison, Claude 3.5 Sonnet is cited at 130, GPT-5 at 150, and Fable 5 at 161 in mid-June.
- Herbie Bradley argues the “human ceiling” is mostly an artifact of benchmark selection: people design new benchmarks around what current models still fail at, so the ceiling keeps moving.
- The same dynamic happened in 2023 and 2024, meaning older benchmarks are already saturated; next year, both benchmarks and model scores are likely to rise together.
Related event: Study Replication Predicts AI to Surpass Human Baseline by 2026(2 posts)→
More from AGI Musings
- First-ever paper on multi-agent swarm safety lands NeurIPS spotlight — LauraRuis · 2026-10-16
- Yoav Artzi on peer review: paper flooding still pays off as review system strains under load — yoavartzi · 2026-10-16
- Letta founder: work in fields defined by problems, not methods — lateinteraction · 2026-10-16
- Why pure mathematicians will soon flood into AI x-risk debates — Dr_Atoosa · 2026-10-16
- RSI began when model outputs augmented next-gen training data, agents just escalated it — akbirthko · 2026-10-16
- Luiza Jarovsky: with AI making careers unpredictable, 'do what you love' is the best advice left — LuizaJarovsky · 2026-10-16