Toby Ord: OpenAI and GDM Data Lag Anthropic's, 80% Horizon Grows to ~25 Minutes
tobyordoxford · x · 2026-11-19
Toby Ord and Ben Todd discuss AI task-horizon reliability data: breakdowns from OpenAI (published twice) and Google DeepMind both show slower growth than Anthropic's reported figures. Short-task data suggests reliability saturates at 80–90%, likely insufficient to close the loop on automating R&D, though the 80% horizon has grown from 15 to 25 minutes. Ord calls on Anthropic to release comparable breakdowns.
Related event: Toby Ord: 80% reliability emerges as bottleneck for AI task length growth(2 posts)→
More from Models
- Gemini 4 Neon places 4th in Sokoban eval as Hydrogen matches Terra at 7.5x cost — banteg · 2026-11-20
- Google confirms its 'most ambitious' Gemini 4 pre-training run is underway — OfficialLoganK · 2026-11-20
- SnakeBench: frontier models still fail to plan more than 2-3 moves ahead — GregKamradt · 2026-11-20
- Datalab shows ParseBench is gameable: a 650M model hit near the top with a few RL steps — VikParuchuri · 2026-11-20
- GPT-6.2 Astra matches predecessor's MazeBench score at $225 vs $840 — patience_cave · 2026-11-20
- MazeBench: GPT-6.2 Sol narrows gap with Astra but still scores less at same price — patience_cave · 2026-11-20