OpenPrediction
METR 50% time horizons of about two weeks around the start of 2028
“I expect doubling times of around 170 days on METR's task suite (or similar tasks) [3] over the next 2 years or so which implies we'll be hitting 2 week 50% reliability horizon lengths around the start of 2028.”
Context: LessWrong post "My AGI timeline updates from GPT-5 (and 2025 so far)" (Aug 2025).
How we judge it
Our interpretation: correct if METR's best published 50% time horizon first reaches about 2 work-weeks (~80 hours) between Jul 2027 and Jun 2028; wrong if it happens before Jul 2027 or is not reached by Jun 2028.
Resolution criteria are our interpretation of the claim, written before grading. Methodology
- Said
- Resolve by
- Due in 630 days
- Verdict
- Open
- Stated probability
- None stated
- Specificity
- 3 of 3 (precise and measurable)
Evidence & grader’s note
Watch: METR measured Claude Opus 4.6 at about 14.5 hours in Feb 2026, which is ahead of a 170-day doubling path.
