Highest-capability Claude alongside Mythos 5, which tops the BenchAlign leaderboard.
Fable 5 arrived in June 2026 as one of Anthropic's two highest-capability models and holds an independent intelligence index of 60 — just behind Opus 5. In practice the two flagships overlap heavily. Opus 5's later arrival and lower input price have made it the default recommendation at the very top, but Fable 5 remains a genuinely frontier-class model with strong agentic and coding behavior and a roughly 1M-token context.
Fable 5 is also the survey's clearest example of how policy now shapes access to frontier models. Anthropic suspended access between June 12 and 30 to comply with a US Department of Commerce export-control directive, then restored it on July 1. The episode is worth noting for anyone building on the model: capability is only half the question, and availability can move for reasons unrelated to the technology.
Output pricing has not been consistently published, which complicates a direct cost comparison against its siblings. For most teams the choice between the two Anthropic flagships will come down to price and whatever availability terms apply rather than to raw capability.
FIELD NOTE — Access suspended Jun 12–30 under a US export-control directive, then fully restored Jul 1.
VISIT OFFICIAL SITE →