Anthropic's Fable 5.2 Is Brilliant — If You Can Afford the Meter

Anthropic released Claude Fable 5.2 and Mythos 5.2 on Monday, promising models that can spin a handful of instructions into working software, design physical objects and tackle research problems that once needed teams of specialists. Output runs up to 30% faster than the previous generation, and token rates dipped to $8 per million input tokens and $40 per million output tokens, down from $10 and $50.

Then there's the meter. AI YouTuber Matthew Berman called Fable 5.2 the best model he has ever used, and burned more than 100 billion tokens testing it. A single Gears of War: E-Day-inspired level took 43 hours and 48 minutes of build time and 8.63 billion tokens, with the running API cost estimate hitting $14,710. A browser-based Excel clone with formulas and pivot tables ran about $11,700 after discounts for cached inputs.

A Fall Guys-style game chewed through more than 9 billion tokens over 71 hours, roughly $13,000. Berman still reported awkward animations, bugs and features that needed testing. Cheaper builds exist: less than $40, or two months of Claude Pro, produced a cute orange cylinder, and a water-balloon physics simulation cost $91.

Anthropic's internal FrontierCode 2.0 evaluation, which tracks the share of proposed coding changes that would be accepted into real open-source projects, put Fable 5.2 at 49.1%, up from 32.4% for Fable 5.1. Anthropic also says Fable helped solve two open theoretical-physics problems involving black holes and design three functioning electronic devices — claims that still need independent scrutiny.

Fable and Mythos share one underlying model, but Fable carries stricter safeguards around dangerous biological and cybersecurity requests, while Mythos is reserved for verified organizations in those fields. Optional fast mode generates output up to 2.5 times faster and doubles rates to $16 and $80. For contrast, Kimi 3.1 Pro costs $2 and $10 per million tokens, and the latest Deepseek model runs $0.6 and $1.98 off-peak. Fable still sits at twice the per-token price of Claude Opus 5.5, which Anthropic recommends for everyday work.

Anthropic's launch offer hands Pro users and standard Team seats one week of Fable access, capped at half their weekly allowance plus a reset; Max, Ultra and eligible premium seats get one to three resets. Pro's promotional access ends November 16, after which Fable returns to paid credits on those plans, and Max and Ultra resets last until March 8, 2027. The sticker shock has precedent: Fable 5 torched some subscribers' quotas in minutes, and Fable 5.1 kept premium pricing when it arrived in September. Last week's rumor mill floated a Fable 5.5 — the actual release is 5.2.