Claude Fable 5.1 and Claude Mythos 5.1
Points and comments are a snapshot, not live.
Anthropic releases Claude Fable 5.1 and Mythos 5.1 with enhanced coding and scientific research capabilities.
Claude Fable 5.1 and Mythos 5.1 are the same model with different safeguards. Fable 5.1 is generally available; Mythos 5.1 is for trusted access programs. Fable 5.1 costs ~25% less than Fable 5 due to reduced cache read pricing. It sets new benchmarks in coding, scientific research, and knowledge work. In tests, it achieved 52.6% on Terminal-Bench-Science 0.1 vs. 24.7% for Fable 5. It designed high-affinity protein binders with ~50% hit rate (typical is 10-15%), created a Venus elevation map, and sped up bio models up to 2.5x via custom GPU kernels.
Enterprise Frontier Safeguards (EFS) give customers full privacy via customer-controlled cloud storage. Safeguards now have ~60% fewer false positives in cybersecurity and 85% fewer in biology. Anti-distillation mechanisms prevent extraction of model capabilities. Mythos 5.1 falls short of the next risk tier in Anthropic's Responsible Scaling Policy.
What commenters are saying
Many commenters report Fable 5 was nearly unusable due to aggressive safeguards that downgraded to Opus on security-adjacent tasks (auth, crypto, reverse engineering). Several users could not complete even basic REST service work without triggering blocks. Some found it improved over time, with fewer downgrades. A few reported success with non-security work like web dev and planning.
The usage reset timing drew complaints from users whose windows were disrupted. Some noted the price cut relies on caching, with real-world cost reductions of ~30%. One user noted quota limits dropped 17% despite the discount, though another pointed out they were temporarily increased in May.