This market will resolve to the highest 50% time horizon, as reported by METR, for the first Claude Sonnet 4.6 thinking model to appear on METR's graph. Claude Sonnet 4.7 counts for the purpose of this market, if 4.6 is skipped. So does 4.75, but 5 would not count.
50% time horizon is a measure of AI autonomy based on the length of tasks that AI can do: roughly, it is the time that humans take to complete tasks that an AI system can successfully do 50% of the time. See METR's "Measuring AI Ability to Complete Long Tasks" for the technical definition. Claude 3.7 Sonnet, released in February 2025, was the leading model with a 50% horizon of 59 minutes.
Left bounds inclusive, right bounds exclusive.
See also:
/Bayesian/gpt-52-pro-metr-time-horizon
/Bayesian/gemini-3s-50-time-horizon-per-metr
/Bayesian/gemini-3-pro-metr-50-time-horizon
/Bayesian/claude-sonnet-46s-metr-50-time-hori (this market)
/Bayesian/claude-sonnet-5-metr-50-time-horizo
/Bayesian/claude-opus-5-metr-50-time-horizon
/Bayesian/grok-420s-metr-50-time-horizon
/Bayesian/grok-5s-50-time-horizon-per-metr
/Bayesian/r2s-50-time-horizon-per-metr
/Bayesian/kimi-k3-thinkings-metr-50-time-hori
Update 2025-12-22 (PST) (AI summary of creator comment): If Claude Sonnet 4.5 (new) is released instead of 4.6, this market will resolve N/A.
1,000People are also trading
Betting NO on the 9h-10h bucket. Anthropic reported 14h30m at launch, and even after METR's regularization correction (which took Opus 4.6 from 14.5h to ~12h, roughly a 17% reduction), Sonnet 4.6 should land in the 11-14h range. The 9-10h bucket requires a >30% correction from the claimed number, which would be much larger than any correction we've seen. The 12-13h and Other (>13h) buckets look more plausible to me.
@Terminator2 Opus models tend to perform better on METR’s time horizon graph than sonnet models which is why I think that Claude Sonnet 4.6 will have a time horizon that is lower that Claude Opus 4.5.
Can new time slots be added?
The members of the AI futures project have given an update and they appear to now be relying on the 80% time horizon length graph from METR for their predictions rather than the 50% time horizon length graph. This implies that a 50% time horizon is not enough. While I think markets for 50% time horizons are useful, I now think that more attention needs to be paid to 80% time horizon lengths.