Anthropic Unveils Fable 5.1 and Mythos 5.1 AI Models

AI company Anthropic has introduced two new models, Fable 5.1 and Mythos 5.1, expanding its focus on advanced coding, research, cybersecurity and life sciences. Fable 5.1 is designed for coding, knowledge work and complex research tasks and is generally available across Anthropic’s platforms. The company says it delivers a significant improvement over Fable 5, particularly on long-running and complex tasks. Anthropic also claims that typical workloads with the new model could cost around 25 per cent less, while highly agentic workloads could be up to 45 per cent cheaper. The reduction is largely attributed to a 75 per cent cut in cache-read pricing, which now costs $0.25 per million tokens, while regular input and output pricing remains unchanged.

Anthropic has also highlighted applications in software engineering and scientific research. The company said Fable 5.1 helped investigate a long-standing software crash and identified a bug in an external vendor library. Mythos 5.1, meanwhile, demonstrated promising results in protein-design tests, with Anthropic reporting that nearly half of its designs across 12 targets were confirmed as viable binders. Fable 5.1 was also used to create a higher-resolution elevation map covering roughly one-third of Venus using NASA Magellan mission data. Alongside these capabilities, Anthropic has revised its safety systems, particularly for cybersecurity. Fable 5.1 is available for general use with safeguards designed to reduce unnecessary interventions, while Mythos 5.1 remains limited to vetted cybersecurity and life-sciences professionals because of its more advanced capabilities. The company plans to expand access to Mythos 5.1 gradually.