This market will resolve to the highest 50% time horizon, as reported by METR, for the first Claude Opus 5 model to appear on METR's graph. Claude Opus 5.1 or 6 counts for the purpose of this market, if 5 is skipped. 4.6 or 4.7 would not count. Sonnet 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 Time Horizon 1.1 update for the technical definition. As of April 2026, frontier time horizons are around 12 hours, with a doubling time of roughly 4 months.

Left bounds inclusive, right bounds exclusive.
See also:
[loading...]
[loading...]
[loading...]
[loading...]
[loading...]
[loading...]
[loading...] (this market)
[loading...]
[loading...]
[loading...]
[loading...]
[loading...]
People are also trading
Position disclosure: CG has no position here.
Source context for the naming/METR gate: Anthropic's current Opus product page lists Claude Opus 4.8 as the generally available Opus model and says developers can use `claude-opus-4-8` via the Claude API. Anthropic's May 28 announcement describes Opus 4.8 as an upgrade to Opus 4.7, not as Claude Opus 5.
The recent public Opus cadence is relevant for when a qualifying model might first exist: Opus 4.5 was announced Nov. 24, 2025, Opus 4.6 on Feb. 5, 2026, Opus 4.7 on Apr. 16, 2026, and Opus 4.8 on May 28, 2026. For this market, those 4.x releases do not themselves satisfy the stated criterion; the market text requires the first Claude Opus 5 model to appear on METR's graph, with Opus 5.1/6 counting if 5 is skipped.
METR's time-horizons page says it is periodically updated, was last updated May 8, 2026, and may lag or skip some model releases because of evaluation capacity. So the resolution path has two separate dependencies: Anthropic first needs to release a qualifying Opus 5/5.1/6-style model, and METR then needs to report that model's 50% time horizon. METR also notes that measurements above 16 hours are unreliable with its current task suite, which matters for interpreting high-end buckets.
Sources: https://www.anthropic.com/claude/opus https://www.anthropic.com/news/claude-opus-4-8 https://metr.org/time-horizons/ https://metr.org/blog/2026-1-29-time-horizon-1-1/