Anthropic's Claude Fable 5.1 & Mythos 5.1 Are Now Official With Serious Upgrades Across Cybersecurity, Research, Coding & More

Anthropic has announced Claude Fable 5.1 and Claude Mythos 5.1, and this launch goes way beyond another benchmark bump. The two models share the same core intelligence, but Mythos gets better safeguards for advanced cybersecurity and life-science research.

Claude Fable 5.1 is coming for serious workloads

Fable 5.1 is Anthropic’s new generally available model for coding, knowledge work, computer use, and long-running agentic tasks. Anthropic says it scored 52.6% on Terminal-Bench-Science, compared with 24.7% for Fable 5.

It also hit 73.4% on CursorBench, while reaching 60.9% on Humanity’s Last Exam with tools. Anthropic says the model can work for hours, recover from failures, verify its output, and continue complex projects with less supervision.

The pricing change is another big part of this launch. Cache reads now cost $0.25 per million tokens, down 75% from Fable 5. Anthropic estimates typical workloads will cost around 25% less, while highly agentic workloads could become roughly 45% cheaper. Fable 5.1 still costs $10 per million input tokens and $50 per million output tokens.

Mythos 5.1 is where things get seriously interesting

Claude Mythos 5.1 is the same underlying model, but Anthropic has built it for vetted cybersecurity and life-science researchers. Access remains limited through trusted programs, with availability currently focused on selected US organizations.

The research results are wild. Anthropic says Mythos 5.1 produced protein binders with nearly 50% viability across 12 targets, compared with typical protein-design hit rates of 10–15%. It also reportedly optimized seven open-source biology models by up to 2.5x, potentially cutting GPU costs by 30–60% in large genomic analyses. That could make Mythos particularly interesting for scientific research.

Anthropic is keeping Mythos behind the gate

That access restriction is deliberate. Anthropic says Mythos 5.1 has its strongest cybersecurity capabilities yet, while its biology abilities remain powerful enough to require tighter controls.

Meanwhile, Fable 5.1 gets more precise safeguards, including permission to identify software vulnerabilities. Anthropic says cybersecurity interventions are around 60% lower than before.

At a time when OpenAI, Google, and Anthropic are racing toward increasingly autonomous AI, this launch shows where the competition may be heading next.

Could Mythos 5.1 become Anthropic’s biggest weapon in the race for AI-powered scientific discovery? Let us know what you think in the comments below.