Anthropic launches Claude Fable 5.1 and Mythos 5.1 models claimed to be ‘world’s most advanced models for coding and knowledge work’

Anthropic has launched two new AI models, Claude Fable 5.1 and Claude Mythos 5.1, with a focus on coding, research and complex knowledge work. The company says the new models can handle longer and harder tasks while also reducing costs for many users. Fable 5.1 is available to everyone, while Mythos 5.1 is limited to approved users working in areas such as cybersecurity and life sciences. The two models are essentially the same model, but come with different safety controls. Anthropic says Mythos 5.1 has more flexible safeguards for vetted researchers and security professionals. Announcing the two models, Anthropic said: “They’re the world’s most advanced models for coding and knowledge work—and their research capabilities offer an early glimpse of how AI models will contribute to scientific progress”.Claude Fable 5.1 brings lower costs and stronger coding performance: AnthropicAnthropic says Claude Fable 5.1 will cost about 25% less than Fable 5 for typical workloads when usage is charged by tokens. For highly agentic tasks, which involve the AI working through many steps and using tools, the savings can reach about 45%.The lower cost comes mainly from a 75% cut in the price of cache reads. Cache reads allow the model to reuse information it has already processed. Anthropic says Fable 5.1 will continue to cost $10 per million input tokens and $50 per million output tokens, while cache reads will cost $0.25 per million tokens.The company also says the model performed better than its predecessor on several tests. On CursorBench 3.2.0, for example, Fable 5.1 scored 73.4%, compared with 70.5% for Fable 5 and 70% for Opus 5.New Claude model aims to help with scientific researchAnthropic is also highlighting the research abilities of its new models. In one test, Claude Fable 5.1 created a new high-resolution elevation map of part of Venus using old radar images from NASA's Magellan mission. Anthropic says the map shows details as small as two to three kilometres, compared with 10 to 20 kilometres in the earlier map.The company also tested Claude Mythos 5.1 for protein design. Anthropic says the model created protein binders that showed strong results in laboratory testing. Across 12 targets, the model achieved a hit rate of nearly 50%, which the company says is much higher than the typical 10% to 15% rate in protein design.Mythos 5.1 was also used to optimize seven open-source biology and genomics models. Anthropic says the models ran up to 2.5 times faster, while estimated GPU costs for some genome-wide analyses fell by 30% to 60%.Claude Mythos 5.1 is restricted to approved usersAnthropic is not making Mythos 5.1 generally available. The model is being offered through trusted access programs for selected cybersecurity professionals and life sciences researchers. The company says these users can access capabilities that are restricted in the standard Fable 5.1 model.For cybersecurity, Anthropic says Fable 5.1 can now help find software vulnerabilities as part of defensive security work. However, tasks such as exploit generation and penetration testing remain restricted.The company also says its updated cybersecurity safeguards produce about 60% fewer false-positive interventions than the previous safeguards.For life sciences, Anthropic has created an access program for professionals using Mythos 5.1 for research and development. The program was developed in partnership with the US government.Anthropic is also introducing Enterprise Frontier Safeguards, or EFS, for enterprise customers. The system is designed to provide privacy similar to a zero-data-retention setup while allowing Anthropic to maintain safeguards against misuse. Under EFS, customer data is stored in cloud infrastructure controlled by the customer rather than Anthropic.The feature will begin rolling out to enterprise customers in phases later this fall.Claude Fable 5.1 and Mythos 5.1 availabilityClaude Fable 5.1 is available immediately across Anthropic's platforms. Developers can also access it through the Claude API, while the model is available through cloud platforms including Amazon Web Services, Google Cloud and Microsoft Azure.Mythos 5.1, meanwhile, remains restricted to vetted users as Anthropic continues to balance its more advanced capabilities with safety concerns.Get the latest technology news and updates. Download the TOI App.