tech
Anthropic releases Claude Fable 5, a 'Mythos-class' AI model with safeguards
Anthropic said it could not yet release the full Claude Mythos 5 model, so instead it created Fable 5 with baked in safeguards.
TL;DR
- Anthropic has launched Claude Fable 5, a 'Mythos-class' AI model with enhanced safeguards.
- Safeguards are triggered for queries related to cybersecurity, biology, and chemistry, routing them to the Opus 4.8 model.
- Fable 5 is initially available to Pro and Max subscribers, with a shift to usage credits starting June 23.
- A version with fewer safeguards, Claude Mythos 5, is undergoing testing with select partners.
- Fable 5 reportedly compresses months of engineering into days, as per Stripe's findings.
- The release coincides with Anthropic's confidential S-1 filing for an IPO, mirroring a similar move by rival OpenAI.