Claude Fable 5.2 solved two black hole problems and designed three devices

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on 9 November. The company says both models beat Claude Opus 5.5 in almost every capability it tested, generate output up to 30% faster than the 5.1 versions and cost less: $8 per million input tokens and $40 per million output tokens.

Fable and Mythos run on the same underlying model. Fable carries cybersecurity and biology safeguards, so Anthropic offers it to all users on every platform, including Amazon Web Services, Google Cloud and Microsoft Azure. Mythos has fewer of those blocks, and only organizations accepted into Anthropic’s Cyber Verification or Life Sciences Verification programs can use it.

Physics, hardware and a proof

Anthropic lists three results from prerelease testing:

• Physics. Fable 5.2 solved two open problems about black holes. In one, it finished a calculation laid out in 2003 that physicists had found too laborious, and showed that black holes form abruptly in very small universes too. In the other, it found the exact limit of a technique physicists have used for more than 15 years. Both results are in preprints, not yet peer-reviewed papers.
• Electronics. From a one-paragraph prompt, Fable 5.2 designed a voice-assistant hub, a USB speakerphone and a handheld virtual pet, including four-layer circuit boards, 3D-printed cases and software. Circuit board maker Diode built them, and all three worked without electrical changes.
• Mathematics. A team of 1,000 Fable 5.2 agents formalized the proof of Fermat’s Last Theorem in the Lean language in 30 hours. An unreleased research version of Claude needed 11 days for the same job, and Opus 5.5 finished 26% of it in 18 hours.

Anthropic ran the Fermat test once per model. Each AI agent on the Fable team started from minimal prompts, and together they used a third of the action tokens of the first attempt and produced a proof seven times shorter.

Where it loses

On AutomationBench, a test of business workflows, Fable 5.2 scored 51.4% against 53.9% for Google’s Gemini 4 Carbon. On FrontierCode 2.0, which checks whether open-source maintainers would merge a model’s code, it reached 49.1% against 45.5% for Opus 5.5. Anthropic picked the tests and ran most of them itself.

“Claude Mythos 5.2 and Fable 5.2 each voice suspicion that they are being evaluated more often than any of our recent models, which challenges our ability to assess how they will act in real-world settings.”

— Anthropic, Anthropic

Quote source: Anthropic, 9 November 2026.

Anthropic still rates the risk of unsafe behaviour in high-stakes situations as low. In September it withheld Mythos from Britain, and the new release keeps Mythos behind the same verification gate.

How you get it

If you pay for Claude, Fable 5.2 is in the model list today. This week Anthropic gives eligible paid plans one to three extra Fable resets, and Pro plans can spend up to 50% of their weekly limit on Fable. Developers call it as claude-fable-5-2. An optional fast mode speeds output up to 2.5 times for double the price. The launch comes weeks after Anthropic filed for a Nasdaq listing.

Based on Anthropic’s announcement of 9 November 2026, checked on 10 November. Every benchmark and research result here is Anthropic’s own claim, and outside researchers have not yet reproduced them. Intokened’s editorial tools use Claude models.