This market will resolve to the highest 50% time horizon, as reported by METR, for the first Claude 4.5 Opus thinking model to appear on METR's graph.
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.
Time horizon could vary based on the set of tasks used to measure it, so this market will be based on the time horizon for the most comprehensive set of tasks reported by METR (as of 2025, largely software and engineering tasks). This will be ambiguous if METR stops publishing time horizons across all of their autonomy tasks and only publishes separate results for different subsets; I might N/A in that scenario.
See also:
/Bayesian/gemini-3s-50-time-horizon-per-metr
/Bayesian/gpt5-pros-50-time-horizon-per-metr
1,000People are also trading
@Amonium at first i thought the same but notice that the >4h one has n=3 so basically meaningless. it seems kind of indicative of low 2h30min ish to me?
@MaxLennartson it means there's only 3 problems in SWE-Bench verified that are estimated to take 4h+ for humans to solve, if I'm understanding it right
@MaxLennartson not exactly. more like: i think opus will have a 2.5ish time horizon based on the scores for human times <4 hours. and i'm mostly dismissing the >4h row because it is not very informative because it has only 3 samples
@jim Would it be possible to split the potential time horizons up? Example 2h-2.5h, 2.5h-3h and so on.
@MaxLennartson it's not possible with how this market is set up. One could create a new market with better buckets though.