Claude Fable 5.2

An independent write-up updated November 10, 2026 summarizes Anthropic's November 9 launch of Claude Fable 5.2 and Claude Mythos 5.2, citing the launch post, Claude Platform release notes, and Anthropic help articles. It confirms the two-name, one-model split: Fable 5.2 is open to all paid Claude plans and API customers with cyber and biology safeguards, while Mythos 5.2 is limited to verified organizations. The guide adds specifications not stated in the launch excerpt: a 1M-token context window, up to 128K output tokens, and adaptive thinking that is always on. It also records the API price cut from $10/$50 per million tokens on the 5.1 models to $8/$40 on 5.2, with an optional fast mode at double the price, and notes Anthropic still recommends starting most work on Claude Opus 5.5.

A Hacker News discussion of Artificial Analysis reporting cites a score of 65 for Claude Fable 5.2, described in the thread title as the highest so far. Community participants focus less on headline scores and more on cost per task, noting that output tokens per task reportedly jumped from about 118k in Fable 5.1 to roughly 274k in Fable 5.2 despite the price cut. One commenter reports Fable 5.2 at a high reasoning effort scoring about 62 at roughly $9.08 per task, versus Claude Opus 5.5 at max effort scoring about 59 at roughly $10.04 per task, arguing the new model extends the cost-performance frontier. These cost figures are reader estimates rather than official measurements.

A third-party benchmark index for Claude Fable 5.2, dated November 10, 2026, places the model near the top of its composite leaderboard: Coding 1 of 307, Reasoning 1 of 407, Vision 1 of 244, Math 2 of 353, and Tool Calling 2 of 233. It also lists a blended price of $9.52 per million tokens and an LLM Stats Score of 61.3. The same page tabulates individual evaluations for the model, including GDPval-AA version 2.1 at an Elo of 1831, AA-Briefcase v1.1 at 1785 with a 66.1% rubric pass rate, and SWE-bench Multilingual at 0.97 with adaptive thinking at max effort. Each figure is attributed back to Anthropic's system card and blog rather than to an independent evaluation run.

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, describing them as the same underlying model with improvements in coding, reasoning, and vision. Fable carries strong cybersecurity and biology safeguards and is available to all users, while Mythos is reserved for organizations approved through Anthropic's expanded Life Sciences and Cyber Verification Programs. Outputs run up to 30% faster than Fable 5.1 and Mythos 5.1, with pricing lowered to $8 per million input tokens and $40 per million output tokens. An optional fast mode generates tokens up to 2.5 times faster for $16 per million input and $80 per million output tokens. Prerelease work included software migrations, three electronic devices, black hole physics results, and formalizing Fermat's Last Theorem in 30 hours.