Anthropic has introduced 'Claude Mythos 5.1 and Fable 5.1', expanding its Claude model lineup with systems designed for advanced coding, research, problem-solving and scientific applications.
The new models arrive after the launch of Fable 5 and sit above Anthropic's existing Haiku, Sonnet and Opus models, which the company categorises as small, medium and large models.
Mythos 5.1 and Fable 5.1 share the same underlying model but have different safety controls and access policies. Fable 5.1 is available to all users, while Mythos 5.1 is currently limited to approved organisations working in areas such as cybersecurity and life sciences.
Fable 5.1 Focuses on Coding and Problem-Solving
Anthropic describes Fable 5.1 as its strongest model so far for coding, research and complex problem-solving tasks.
In the Terminal-Bench Science 0.1 test, Fable 5.1 achieved a score of 52.6%, compared with 24.7% for Fable 5, 29.0% for Opus 5 and 22.4% for GPT-5.6.
The model also scored 55.8% on Terminal-Bench 4.0. Mythos 5.1 performed even better in the same benchmark, reaching 60.9%. By comparison, Fable 5 scored 42.0%, Opus 5 achieved 52.3%, and GPT-5.6 scored 37.3%.
Anthropic has also improved the models' cybersecurity capabilities. The company says its updated safety checks have reduced cybersecurity false alarms by 60%. Fable 5.1 can identify software security vulnerabilities, although it is not designed to exploit them.
Lower Costs for Fable 5.1
Anthropic has also reduced the cost of using Fable 5.1. The company has cut the price of cache reads by 75% to $0.25 per million tokens. Cache reads let models reuse information they have already processed instead of reprocessing it.
According to Anthropic, Fable 5.1 costs around 25% less than Fable 5 during typical usage and can be up to 45% cheaper for demanding workloads such as complex coding.
Standard input and output pricing remains unchanged at $10 and $50 per million tokens, respectively.
Mythos 5.1 Targets Scientific Research
Mythos 5.1 includes additional capabilities for scientific research. Anthropic says the model has demonstrated progress in areas including drug discovery and biology.
During drug-discovery tests involving 12 targets, Mythos 5.1 designed molecules for potential medicines and achieved a success rate of nearly 50%. Anthropic compared this with a typical industry success rate of around 10% to 15%.
The model has also been used in biology research programmes, where Anthropic says it helped teams complete work up to 2.5 times faster while reducing computing costs by between 30% and 60%.
Another application involved mapping part of Venus using older NASA radar data. Fable 5.1 produced a detailed map that Anthropic says contained more detail than previous maps.
New Enterprise Data Controls
Anthropic is also preparing Enterprise Frontier Safeguards (EFS), a system intended to give business customers greater control over sensitive information.
The system is designed to let organisations keep their data on their own servers instead of Anthropic's infrastructure, while retaining safeguards intended to prevent misuse.
EFS is expected to be introduced in stages from autumn 2026 across services and platforms including:
Claude Code
Claude Enterprise
Amazon Bedrock
Google's agent platform
Microsoft Foundry
Until the system becomes available, eligible customers can use Fable 5.1 with no data stored, according to Anthropic.
Safety Testing Before Release
Both Fable 5.1 and Mythos 5.1 underwent safety evaluations covering potential risks involving chemical, biological and cybersecurity threats before their release.
Anthropic says the assessments did not identify serious ways to bypass the models' safety protections.
The company has also placed additional restrictions on Mythos 5.1 because of its capabilities in sensitive research areas.
Availability
Fable 5.1 is available through Anthropic's API, as well as Amazon Web Services, Google Cloud and Microsoft Azure.
Mythos 5.1 currently has more restricted access and is available only to a limited number of approved organisations in the US, particularly those working in cybersecurity and life sciences.
Anthropic says it is working with the US government to expand access to additional users, including organisations outside the US.
