This market will resolve to the highest 50% time horizon, as reported by METR, for the first Claude Mythos model to appear on METR's graph.
'Mythos' includes any Claude model that sits above 'Opus' in Anthropic's naming scheme, even if Anthropic ends up calling it something other than Mythos.
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.
Update 2026-03-29 (PST) (AI summary of creator comment): Resolves N/A if no qualifying model is released by September 30th (2026).
1,000People are also trading
This seems very low to me? 43% of 30-50 puts that at 4% per 2h increment, on average. Meanwhile, converting from AECI puts Mythos at around 40 hours from my calculation--so I'd have the mode much higher. AECI is strictly a superset of publicly available metrics right so it's our most trusted source as I see it.
I know METR will be degenerate (massively noisy) at 40h, but is there reason to expect it to be downward biased?
@MaxLennartson I'm not expecting it. Maybe they'll eval a later version? Otherwise market will resolve N/A if no qualifying model is released so that METR can evaluate it by September 30th