Markets live
Data: CoinGecko

Claude Fable 5.2: what’s new, how much it costs and how to get it

Anthropic’s Claude Fable 5.2 is faster and 20% cheaper than Fable 5.1 and tops a key AI leaderboard. Prices, plans, Mythos 5.2 and how it compares with Gemini 4 Carbon.

Anthropic CEO Dario Amodei speaking on stage; Anthropic released Claude Fable 5.2 on Nov. 9, 2026
Anthropic chief executive Dario Amodei at TechCrunch Disrupt in San Francisco, 2023. Photo: TechCrunch, CC BY 2.0, via Wikimedia Commons, cropped.

Anthropic released Claude Fable 5.2, its most capable AI model for general use, on Nov. 9, 2026. The company says it is faster and cheaper than Fable 5.1 and “a step above” Claude Opus 5.5 in almost every skill it tested. The release comes days before Anthropic’s planned stock market listing.

The essentials

  • Claude Fable 5.2 launched on Nov. 9, 2026, for Pro, Max, Ultra, Team and Enterprise users, and through Anthropic’s API and the AWS, Google Cloud and Microsoft clouds.
  • API prices fall to $8 per million input tokens and $40 per million output tokens, from $10 and $50 for Fable 5.1, according to Anthropic.
  • A twin model, Claude Mythos 5.2, has fewer safety blocks on cybersecurity and biology and is only for vetted organisations.

The launch puts Anthropic back at the top of one widely watched leaderboard, less than two weeks after Google claimed the lead with Gemini 4 Carbon. For businesses, the bigger story may be price: the most capable Claude model now costs 20% less to use. Anthropic, which is preparing to go public, competes directly with Google and OpenAI for corporate AI spending.

What’s new in Claude Fable 5.2?

Fable 5.2 is Anthropic’s top model for hard, long-running work such as complex coding, research and multi-step “agent” tasks, where the AI uses tools and takes actions on its own. In its announcement, Anthropic said the model produces answers up to 30% faster than Fable 5.1 and Mythos 5.1.

Other changes include:

  • Fast mode: an option that generates text up to 2.5 times faster, at double the standard price.
  • Medium effort by default: in the Claude apps and Claude Code, the model now spends a medium amount of “thinking” on each task unless users choose more.
  • Fewer false refusals: Anthropic says its safety filters blocked about 55% fewer requests than Fable 5.1 in early testing.
  • Stronger coding scores: Fable 5.2 scored 69.9% on the FrontierSWE v2 software-engineering test, against 62.3% for Opus 5.5, according to Anthropic’s figures.

Anthropic also said a team of about 1,000 Fable 5.2 agents formally verified Fermat’s Last Theorem in 30 hours, a claim that has not been independently checked.

How much does Claude Fable 5.2 cost?

For developers, Claude Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1, according to Anthropic’s product page. A token is a small chunk of text, roughly three-quarters of a word. Fast mode costs $16 and $80, and reading from the cache costs $0.20 per million tokens.

In the Claude apps, Fable 5.2 comes with paid plans rather than a separate fee. During launch week, Anthropic gave eligible subscribers usage-limit resets and allowed Pro users and standard Team seats to spend up to 50% of their weekly limit on Fable, 9to5Mac reported.

Who can use it?

Fable 5.2 is available to Claude Pro, Max, Ultra, Team and Enterprise subscribers, and to developers under the model ID claude-fable-5-2. It is also offered through Amazon Web Services, Google Cloud and Microsoft’s cloud. Free users do not get it.

There is one catch for companies. By default, using Fable models requires Anthropic to keep prompts and responses for 30 days for safety monitoring. Eligible customers can use them with zero data retention, and a system that runs the monitoring inside a customer’s own infrastructure is being tested with Bridgewater, Salesforce and Stripe.

What is Claude Mythos 5.2?

Claude Mythos 5.2 is the same underlying model as Fable 5.2, without some of its cybersecurity and biology safeguards. Anthropic only offers it to vetted organisations through its Cyber Verification Program and Life Sciences Verification Program, for uses such as finding software vulnerabilities or biomedical research.

On Fable 5.2, requests flagged as high-risk in those two areas are handed to an older Claude model, which answers instead. Anthropic says this lets it release the stronger model to everyone while limiting misuse.

How does Fable 5.2 compare with Gemini 4 Carbon?

Claude Fable 5.2 beats Google’s Gemini 4 Carbon on most of the tests Anthropic published, but not all of them, and Gemini is far cheaper. On the independent Artificial Analysis Intelligence Index, Fable 5.2 at maximum effort scores 65, against 60 for Gemini 4 Carbon and 59 for Opus 5.5.

TestClaude Fable 5.2Claude Opus 5.5Gemini 4 Carbon
Artificial Analysis Intelligence Index655960
FrontierSWE v2 (coding)69.9%62.3%61.2%
OSWorld 2.1 (using a computer)89.3%87.2%81.5%
AutomationBench (business workflows)51.4%42.5%53.9%
API price, input / output per million tokens$8 / $40n/a$2 / $10 until Nov. 30, then $4 / $20
Benchmark scores from Anthropic’s Nov. 9, 2026, announcement (each model at its highest available effort), except the Intelligence Index, from Artificial Analysis as of Nov. 13, 2026. Gemini prices from Google.

The results are not a clean sweep. Gemini 4 Carbon scored higher on AutomationBench, which tests everyday business tasks, and Opus 5.5 edged Fable 5.2 on one office-work rating. Google made Gemini 4 Carbon widely available on Nov. 10, and its introductory price is a quarter of Fable’s.

Anthropic itself still recommends Opus 5.5 as the everyday model for most coding and knowledge work, and suggests moving up to Fable only for the hardest jobs. For more on the companies behind the AI boom, see our coverage of Nvidia’s leadership change.

Sources

Related, Nov. 14, 2026: Four nonprofits have sued Anthropic and four other firms to freeze AI development.

Related, Nov. 14, 2026: How ChatGPT Wallet works, who qualifies and which cards it accepts.

Related, Nov. 15, 2026: Meta fined 805 million euros by Italy’s data protection authority: what the Garante found.

Related, Nov. 16, 2026: Anthropic plans to price its IPO on Nov. 18 and start trading on Nasdaq as ANT on Nov. 19. See the Anthropic IPO date and range.

Martin Mwenda Avatar

This article is for information only and is not financial advice. Read our editorial standards and ethics policy.