Anthropic Cuts Claude Fable Price 20%, Says It Can Match Opus For A Tenth Of The Cost

Anthropic cut the price of its most powerful AI by 20% with Claude Fable 5.2 and Claude Mythos 5.2, and says the new model can match its everyday Opus 5.5 on some tests for less than a tenth of the cost per job.

Key Takeaways

Fable 5.2 and Mythos 5.2 cost $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1
Per token, Fable 5.2 still costs twice as much as Claude Opus 5.5, which is priced at $4 and $20
Anthropic says Fable 5.2 matches Opus 5.5’s best scores on several tests at half, 40% or under a tenth of the cost per task
A fast mode runs up to 2.5 times quicker but doubles the price to $16 and $80

Anthropic announced the models on Monday. A token is a small chunk of text, roughly three quarters of a word, and it is how AI companies charge for their models.

The New Price Tag

Price per million tokens	Input	Output
Fable 5.1 (old)	$10	$50
Fable 5.2 and Mythos 5.2	$8	$40
Fast mode, up to 2.5x quicker	$16	$80
Opus 5.5, for comparison	$4	$20

Prices from Anthropic’s pricing page, Nov. 9

Reading saved, cached text also got cheaper, falling to $0.20 per million tokens from $0.25. Customers who need their work processed only in the US pay 1.1 times the normal price, and batch jobs that can wait still get a 50% discount.

Pricier Per Word, Cheaper Per Job

The bigger cost story is how many tokens Fable 5.2 needs. Anthropic says it finishes hard tasks with fewer tokens than before, so a job can cost less overall even though each token costs more than with Opus.

Test	Fable 5.2 Result	Cost Per Task Vs Opus 5.5
Expert exam, Humanity’s Last Exam	Matches Opus 5.5’s top score	Under one-tenth
Reading charts, Chartography	66.4%, ahead of Opus 5.5’s 64.4%	Under one-tenth
Research agents, WANDR	Matches Opus 5.5’s top score	About 40%
Coding, FrontierCode 2.0	45.3% vs Opus 5.5’s best 45.5%	About half

Anthropic’s own test results, published Nov. 9

At its highest setting, Fable 5.2 scored 49.1% on the FrontierCode coding test, which Anthropic says is the best of any released model it has tested, ahead of OpenAI’s GPT-6 Astra at 42.8%.

Also Read: Apiiro Finds 2,539 Secrets Leaked By Claude Code Users On GitHub

What You Get For The Money

Anthropic said that during testing the models solved two open problems in black hole physics, designed three working electronic devices from scratch and formalized the proof of Fermat’s Last Theorem in 30 hours. They also write their answers up to 30% faster than the previous versions.

Fable and Mythos are the same underlying model at the same price. Fable carries strong cybersecurity and biology safeguards so anyone on a paid Claude plan can use it, while Mythos is only open to vetted cybersecurity and life sciences groups.

The Catch

The cost savings come from Anthropic’s own tests, and real bills depend on the job. Anthropic itself still tells most users to start with the cheaper Opus 5.5 for everyday coding and office work, and to step up to Fable only for the hardest problems. Using Fable also requires 30-day data retention for safety monitoring by default.