OpenAI's own report shows models' 80% time horizon on R&D is just 15 minutes
tobyordoxford · x · 2026-10-22
Toby Ord highlighted @ramez's analysis of recursive self-improvement, pointing to OpenAI's own report on research automation: models' 80% time horizon on R&D work is only 15 minutes. The accompanying OpenAI chart shows longer tasks need more interventions — 86% of sub-15-minute tasks succeed with zero interventions, but at the 64-128h range, failures hit 26% and zero-intervention success drops to 16%, underscoring the gap between the automated-AI-researcher narrative and reality.
More from Models
- Baseten releases open-source decision models elephant-1 and mammoth-1, free for a week — baseten · 2026-10-22
- Recursive SI open-sources Flint, a local coding model hitting 125+ tokens/sec on MacBook Pro — CodeByPoonam · 2026-10-22
- Recursive's Flint: a 40B local coding model built in 4 months, 125+ tok/s on a MacBook — jeffclune · 2026-10-22
- NVIDIA Nemotron demos go live — mervenoyann · 2026-10-22
- vLLM adds day-0 support for Flint: 500 TPS on a single GB200 — _rockt · 2026-10-22
- WIRED: Recursive's Flint shows AI-built AI is getting closer to real recursive self-improvement — cong_ml · 2026-10-22