Anthropic Unveils Claude Fable 5 And Claude Mythos 5, Its Most Powerful AI Models Yet
Anthropic has unveiled Claude Fable 5, its most capable AI model yet, bringing major advances in coding, research and long-duration autonomous tasks. The company also introduced Claude Mythos 5, a more powerful version with fewer safeguards that will initially be restricted to cybersecurity partners and researchers due to its advanced capabilities.
Anthropic has announced the launch of Claude Fable 5, describing it as the most capable AI model the company has ever made available for general use. Alongside it, the company also introduced Claude Mythos 5, a more powerful version that will initially be limited to cybersecurity partners and select researchers due to safety concerns.
According to Anthropic, Claude Fable 5 delivers state-of-the-art performance across software engineering, knowledge work, scientific research, vision tasks and long-duration autonomous workflows. The company claims the model can handle complex projects over extended periods, outperforming previous Claude models in coding, reasoning and memory-intensive tasks. Anthropic says Fable 5 can remain focused across millions of tokens and has demonstrated significant gains in analytical and research-oriented applications.
Despite its capabilities, Anthropic says the model’s advanced cybersecurity and scientific knowledge creates new risks. To address this, Fable 5 includes a new safeguard system that automatically routes certain cybersecurity, biology, chemistry and model-distillation related requests to the less capable Claude Opus 4.8 model instead. Anthropic says these safeguards are designed to prevent misuse while still allowing most users to access Fable 5’s capabilities, with more than 95 percent of sessions reportedly requiring no fallback.
The company also revealed Claude Mythos 5, which uses the same underlying model as Fable 5 but with some safeguards removed. Anthropic says Mythos 5 offers the strongest cybersecurity capabilities of any AI model it has created and will initially be deployed through Project Glasswing, a program focused on supporting cyber defenders and critical infrastructure providers. Access is expected to expand gradually through a trusted access program.
Anthropic highlighted several research achievements enabled by the new models, including autonomous genomics research, protein design work and the generation of novel scientific hypotheses. The company also said internal teams observed significant acceleration in drug discovery-related workflows when using Mythos 5.
Claude Fable 5 is available immediately, while Claude Mythos 5 remains restricted. Both models are priced at $10 per million input tokens and $50 per million output tokens. Anthropic says the launch represents another step toward making advanced AI systems broadly available while balancing safety and security concerns.