Anthropic cut Claude Fable’s price 20% on Monday, and that’s the least interesting number in the release.
The headline math is simple. Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. Mythos 5.2 launched at the same prices, but it stays limited to vetted cybersecurity and life sciences organizations. Anthropic also says both models generate output up to 30% faster.
Price per token is only half of a bill. The other half is how many tokens a model burns to finish the job, and that’s where this release gets interesting. On FrontierCode 2.0, an agentic coding benchmark, Anthropic says Fable 5.2 at High effort scores 45.3%. That matches Opus 5.5’s best score (45.5%) for roughly half the cost per task. Fable 5.1’s best was 32.4%.
The effort dial is now the biggest line item. VentureBeat, reading Anthropic’s cost charts, puts High at $6.72 a task and Xhigh at $14.19. Xhigh buys 3.8 more points (49.1%) for more than double the money. For most of what I’d hand an agent, that isn’t worth it.
Two cautions. Every score here is Anthropic’s own, and the cost charts don’t label their values. And Dianne Penn, Anthropic’s head of product management for research, told CNBC that Fable 5.2 is for “power users or power use cases,” while Opus 5.5 is still the better fit for daily tasks. A bigger model isn’t a better default.
I run local models on an old Radeon RX 580, where cost per task is measured in minutes and patience. It turns out that’s the right unit for a frontier API, too.
So run your own tasks on claude-fable-5-2 at Medium and High, and save Xhigh for the problems that earn it. The direction is the good news: the frontier is getting better and cheaper at the same time. My bet is that the teams who benefit most will be the ones who measure cost per task instead of trusting the price sheet.
Sources
- Introducing Claude Fable 5.2 and Claude Mythos 5.2, Anthropic, Nov 9, 2026
- Claude Fable, Anthropic (pricing, availability, model ID), Nov 2026
- Anthropic releases Claude Fable 5.2 and Mythos 5.2, pairing new, record-high benchmarks with 20% API cost reduction, VentureBeat (Carl Franzen), Nov 9, 2026
- Anthropic releases Claude Fable 5.2 and Claude Mythos 5.2 ahead of IPO, CNBC (Ashley Capoot), Nov 9, 2026
- Anthropic launches Claude Fable 5.2 and Mythos 5.2, and cuts Fable’s price, The Next Web (Ana Maria Constantin), Nov 9, 2026