The two models share the same underlying system, but Anthropic has separated access according to risk. Fable 5.1 carries tighter protections intended for broad commercial use, while Mythos 5.1 offers more permissive settings for approved organisations working in areas where the company says advanced capabilities could be beneficial but also misused.
Fable 5.1 is available to Pro, Max, Team and Enterprise users and through Anthropic’s developer platform, cloud marketplaces, Amazon Web Services, Google Cloud and Microsoft Foundry. Anthropic prices the model at $10 per million input tokens and $50 per million output tokens, unchanged from Fable 5.
The main pricing change is for cached input. Cache reads now cost $0.25 per million tokens, down from $1, a 75 per cent reduction. Anthropic estimates the change will lower the effective cost of typical token-billed workloads by about 25 per cent and reduce costs by as much as roughly 45 per cent for highly agentic tasks that repeatedly reuse stored context.
Such workloads can involve software repositories, system instructions, tool definitions, documents and lengthy conversation histories that agents revisit during extended tasks. The lower cache-read charge is therefore aimed particularly at long-running coding, research and enterprise workflows, where repeated context retrieval can represent a significant share of total usage.
Anthropic said Fable 5.1 improves on its predecessor in coding, knowledge work and sustained problem-solving, while offering lower-cost performance at low and medium effort settings. The model has a one-million-token context window and supports outputs of up to 128,000 tokens, with adaptive reasoning enabled.
The company also introduced Enterprise Frontier Safeguards, or EFS, a system designed to combine misuse detection with privacy protections comparable to zero data retention. Under the approach, customer information is stored in cloud infrastructure controlled by the customer rather than Anthropic, while human review is intended to remain with the customer by default.
Anthropic said EFS was developed with more than 100 organisations across financial services, healthcare, manufacturing, telecommunications, law, retail and the public sector, alongside Amazon Web Services, Google Cloud and Microsoft Azure. The system is due to roll out in phases beginning this autumn across Claude Code, Claude Enterprise, the Claude Platform and supported cloud services.
The company has also refined Fable 5.1’s safety controls to reduce unnecessary interventions. It said its updated cybersecurity safeguards should produce around 60 per cent fewer interventions per Claude Code session than the safeguards used with Fable 5. Fable 5.1 can now be used to identify software vulnerabilities, although exploit generation, penetration testing and binary-based vulnerability scanning remain restricted or redirected to other Claude models.
Biology protections have also been adjusted. Anthropic said safeguards for benign elementary biology and medical requests now intervene 85 per cent less often than those introduced with Fable 5, while life-sciences research and development queries remain subject to tighter controls.
Mythos 5.1 is intended for users whose legitimate work requires fewer restrictions. Anthropic is making it available through trusted-access programmes for vetted cyberdefenders and life scientists. Its Life Sciences Verification Program is beginning as an invitation-only beta, while a Cyber Verification Program is expected to add Mythos-class access for defensive security work.
Access to Mythos 5.1 is currently limited to a set of organisations in the United States. The model uses the same one-million-token context window and pricing structure as Fable 5.1, but Anthropic requires a 30-day data-retention period by default for safety monitoring.
Anthropic also highlighted research tests intended to demonstrate the model family’s scientific reach. It said Mythos 5.1 designed high-affinity protein binders that were experimentally validated, while Fable 5.1 produced a higher-resolution elevation map covering about one-third of Venus from archival radar data. Anthropic also said Mythos 5.1 accelerated seven open-source biology models by as much as 2.5 times through customised GPU kernels, potentially reducing computing costs for large genomic analyses.
Follow Arabian Post
Select Arabian Post as your preferred source on Google and MSN News for trusted business news and Arab politics and updates.
