Will a model achieve a METR 50% time-horizon of 4+ hours by the end of 2025?
22
100Ṁ1941Dec 31
23%
chance
10
1H
6H
1D
1W
1M
ALL
METR's 50%-time horizon doubling time trendline from more recent models indicates that a model released towards the end of 2025 will likely have a time horizon of over 4 hours. Will this come true?
See links for more details:
https://www.alignmentforum.org/posts/SuvWoLaGiNjPDcA7d/metr-s-evaluation-of-gpt-5
https://www.lesswrong.com/posts/deesrjitvXM4xYGZd/metr-measuring-ai-ability-to-complete-long-tasks
This question is managed and resolved by Manifold.
Get 
1,000 to start trading!
1,000
People are also trading
Related questions
Will a publicly known AI model achieve an 80% time horizon that is an 1 hour and 30 minutes by September 2026?
46% chance
Will any AI model achieve > 40% on Frontier Math before 2026?
30% chance
Will the METR long-horizons have a >6 month doubling time for at least a 4 month period before 2026?
87% chance
Will a Google model lead METR's task duration chart by EOY?
35% chance
R2's METR 50% time horizon
Will I be able to complete a 10+ hour trip via self-driving car by end of 2025?
3% chance
Will a single model achieve superhuman performance on all OpenAI gym environments by 2025?
22% chance
Model with METR time horizon ≥8 hours released in 2026?
53% chance
Will a single model achieve superhuman performance on all Atari environments by 2025?
14% chance
Will a lab train a >=1e26 FLOP state space model before the end of 2025?
11% chance