Anthropic releases Mythos 5.2 and Fable 5.2 amid RSI concerns
Original title: Anthropic releases new Mythos and Fable models amid RSI concerns
Fable input pricing falls to $8 per million tokens, with 30% faster responses and enterprise zero-data-retention support.
Anthropic launched Mythos 5.2 and Fable 5.2, citing gains in coding, reasoning, vision and clearer replies. Fable input tokens drop from $10 to $8 per million, undercutting GPT-6 Astra, and responses are 30% faster; faster serving costs double. Fable 5.2 supports zero data retention through Enterprise Frontier Safeguards. A METR report suggested internal models sped Anthropic's progress by roughly 2x, with uncertainties remaining. Anthropic says Mythos cannot act covertly against users, while noting stronger covert scores when reasoning is hidden.
Unconfirmed: The roughly 2x METR figure remains uncertain, physics and theorem claims are Anthropic's, and the PSI model is unconfirmed.
- Both models voice suspicion that they are being evaluated more often than recent models.
- Anthropic says Fable 5.2 solved two theoretical-physics problems and formalized Fermat's Last Theorem.
- Physical Superintelligence credited a non-public Anthropic model for an Aaronson-Ambainis proof.
- Anthropic did not confirm whether that model was Fable 5.2.
- Vendors
- Anthropic
- Organizations
- METRPhysical Superintelligence
- AI models
- Mythos 5.2Fable 5.2Claude Mythos 5.2
From the original article
2 min read at techcrunch.com
Anthropic says the new models "show improvements in areas like coding, reasoning, and vision," as well as in communicating more clearly with users.
By Russell Brandom · techcrunch.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.