cpaua
·1 min2

Claude Fable 5.1: Cheaper, Safer Enterprise AI With Science Boosts

Claude Fable 5.1: Cheaper, Safer Enterprise AI With Science Boosts
photo_3890.jpgphoto_3891.jpgphoto_3892.jpg

Released Claude: Fable 5.1. In addition to metric gains (on several fresh and/or closed benchmarks — substantial ones, see the image):
— reading from cache now costs 75% less. Overall this will lead to a 25–45% reduction in price per task solved, depending on the scenario.
— previously Anthropic stored your requests in Fable/Mythos for up to 30 days to analyze them and detect attacker patterns. Many enterprise customers prefer Zero Data Retention, where nothing remains on the server after their requests — which creates a mismatch. For Fable 5.1, Enterprise Frontier Safeguards will be provided — Anthropic will deploy their classifiers and analyzers on your infrastructure and will not store anything on their side. This may lead to broad adoption in the business sector (since so far it has been limping).
— fine-tuned for scientific tasks, especially in biology

In honor of the release, they also lowered the limits!

Share:
Author
cpaua

VibeCode blog admin. Writing about vibe coding, AI and open source.

Comments

To leave a comment, log in or sign up
Loading...