Anthropic’s Fable 5.2 has entered gray testing in Claude Code, Chat, and Cowork, with some requests originally calling Fable 5.1 quietly redirected to Fable 5.2 in the background, according to online reports. Users say the update is so substantial that Anthropic skipped the “5.1” designation entirely. Under maximum reasoning mode, Fable 5.2’s actual performance surpasses GPT-6 Astra, based on extensive overnight testing. AI testing blogger @notjazii found Fable 5.1 was routed to the new version and compared Fable 5.2 with GPT-6 Astra using the same prompt in high reasoning mode, concluding Fable 5.2’s output was clearly superior, while noting it was slower and more expensive due to deeper “slow thinking.” User @Bhavani_000007 ran a “rocket test” with Fable 5.2 and Opus 5.2, also in gray testing, and called the output “eerily realistic.” Opus 5.2, internal code name Opus-Next, has also been announced and may be released soon, possibly even today, after experiencing a “launch-unplug-relaunch” scenario yesterday. This week, none of the big tech companies in Silicon Valley released models.