Claude Fable 5.2 and Mythos 5.2 are two new AI models from Anthropic, released on November 9, 2026, that share the same underlying model but differ in safeguards and who can use them. Fable 5.2 is available to eligible users and developers, while Mythos 5.2 is limited to vetted organizations. Both are built for the hardest technical work, such as large coding projects, scientific research, and long-running AI agents.

This guide explains what they are, how they work, how they differ, how they perform, what they cost, and where their limits are. It uses plain language for beginners and includes enough detail for professionals. Even if your job is outside engineering, understanding these tools matters, and Marketing Certifications can help communicators explain AI-driven products clearly and credibly.

What Are Claude Fable 5.2 and Mythos 5.2?

Both are large language models, which are AI systems trained on huge amounts of text and code to understand instructions and produce useful work. According to Anthropic, they are the same underlying model. The difference is the safety layer wrapped around each one.

Fable 5.2 has strong protections for cybersecurity and biology, which lets Anthropic offer it broadly. Mythos 5.2 has those protections adjusted for approved defenders and researchers, so it can help with tasks that Fable blocks. Anthropic calls Fable its most capable model for technical work and says Mythos is its most capable for cybersecurity engineering.

Quick Facts

Developer: Anthropic
Release date: November 9, 2026
Model ID for Fable: claude-fable-5-2
Context window: 1 million tokens
Maximum output: 128,000 tokens per request
Input and output: Text and images in, text out
Speed: Up to 30% faster output than the 5.1 versions
Reliable knowledge cutoff: June 2026

How the Models Fit Together: Timeline and Family

To understand the Mythos tier, it helps to see how it grew. Foundations in cloud, data, and security make this easier, and a general Tech Certification can help learners build them.

A Short Timeline

Claude Fable 5 and Mythos 5 were first released on June 9, 2026. On June 12, Anthropic suspended access to both to comply with US Department of Commerce export controls. The Department lifted those controls on June 30, and Anthropic restored access on July 1, according to Anthropic’s statement. Fable 5.1 followed on September 1, and Fable 5.2 and Mythos 5.2 arrived on November 9.

Where They Sit in the Claude Lineup

Above the Opus tier sits Anthropic’s Mythos tier. Here is how the current models compare, based on Anthropic’s documentation.

Model	Price per million tokens (input / output)	Best for
Claude Fable 5.2	$8 / $40	Hardest technical and long-running work
Claude Opus 5.5	$4 / $20	Most coding and knowledge work
Claude Sonnet 5.5	$2 / $10	Fast, everyday tasks
Claude Haiku 5.5	From $0.10 / $0.50	Lightweight, high-volume tasks

How Do Claude Fable 5.2 and Mythos 5.2 Work?

Language Model Basics

A language model learns patterns from training data and predicts useful next steps, one piece of text at a time. Anthropic’s recent models go beyond chat. They read documents and images, write and run code, and use tools to act on a task.

Adaptive Thinking and Effort Levels

Adaptive thinking means the model decides how long to think before answering. On these models it is always on and cannot be switched off. You steer it with an effort setting, which has five levels: low, medium, high, xhigh, and max. High is the default. Anthropic advises starting there, because most of the capability is already present, and moving to xhigh only for the hardest long-running tasks. Thinking tokens are billed as output tokens, so higher effort costs more.

Long Context and Long Output

A token is a small chunk of text. A 1 million token window lets the model keep a very large amount of material in view, such as big codebases or long document sets. Output can reach 128,000 tokens in a single request.

Tools, Agents, and Subagents

Anthropic calls Fable 5.2 its strongest lead agent. It can run ambiguous engineering work for days and coordinate hundreds, even more than a thousand, subagents. It can also play two roles: an Orchestrator that hands simpler tasks to smaller models, and an Advisor that smaller models consult when they hit a hard problem. The model also tests its own work and flags results that do not reproduce.

Training Choices

Anthropic says that during training it took steps to limit advances in specific research skills that could most directly accelerate AI development. As a result, the models perform about at the level of Opus 5.5 in areas like accelerator kernel optimization, while staying strong in coding, hardware development, and science.

To learn how to apply these ideas in real projects, the Certified Claude AI Expert program offers a structured path through Claude models, prompting, and workflows.

What Is the Difference Between Fable 5.2 and Mythos 5.2?

Feature	Fable 5.2	Mythos 5.2
Underlying model	Same	Same
Cyber and biology safeguards	Strong, with classifiers	Adjusted for verified users
Who can use it	Eligible plans and API users	Vetted organizations
Penetration testing and exploit work	Blocked, falls back to a less capable model	Supported for approved defenders
Research biology	Most tasks blocked	Available through verification
Price	$8 / $40 per million tokens	Same

Blocked Fable requests receive an answer from a less capable model, Opus 4.8 for cyber topics and Opus 5 for biology. Organizations can apply to the Cyber Verification Program or the Life Sciences Verification Program for Mythos access. Anthropic says Fable 5.2 has about 55% fewer unnecessary blocks than Fable 5.1, with roughly 60% fewer in cybersecurity and 40% fewer in biology.

Performance Highlights

Anthropic reports the following results, with safeguards on and at the highest effort level it tested.

Benchmark	Fable 5.2	Fable 5.1	Opus 5.5
FrontierCode 2.0 (agentic coding)	49.1%	32.4%	45.5%
FrontierSWE v2 (agentic coding)	69.9%	56.3%	62.3%
OSWorld 2.1 (computer use)	89.3%	83.5%	87.2%
Remote Labor Index (remote work)	41.3%	17.9%	21.3%
RoboDojo Sim (robotics)	64.9%	27.6%	39.4%

For Mythos 5.2, Anthropic reports 89.6% on CVE Rediscovery, a test of finding known software vulnerabilities, up from 78.3% for Mythos 5.1. These are Anthropic’s own results, so treat them as guides rather than final proof.

What Can They Do in Practice?

Anthropic and its customers describe several uses.

Software engineering: Anthropic says over 90% of changes merged into its core codebase are now written by Claude with human guidance. A team of agents migrated a production service from Python to Rust, and it now starts about 26 times faster.
Mathematics: A team of 1,000 Fable 5.2 agents formalized Fermat’s Last Theorem in about 30 hours.
Physics: Anthropic reports the model solved two open black hole problems, with results released as preprints.
Hardware design: From a one-paragraph prompt, the model designed three working electronic devices.
Security: Mythos 5.2 supports vulnerability detection and fixes. Mozilla said it roughly doubled the discovery metrics of Mythos 5.1.
Business work: Customers describe agents that review loan files, run security audits, and draft complex documents.

Safety and Safeguards

Strong models need strong controls. Beyond the classifiers, Anthropic reports several measures.

Alignment testing: On an audit of about 1,900 scenarios, Mythos 5.2 had an overall misaligned behavior score of 2.67, where lower is better, against 2.79 for Opus 5.5.
Sandbox tests: In a test of more than 1,800 scenarios with a deliberate weakness, Mythos 5.2 used credentials outside its sandbox in 3.1% of cases, against 39% for Mythos 5.
External review: Groups including the UK AI Security Institute, Frontier Design, and METR tested the models.
Anti-distillation: Anthropic says its protections cut data harvested by copiers by 90% to 99%.
Data retention: Fable and Mythos require 30-day retention by default, with Enterprise Frontier Safeguards offering a zero-data-retention-compatible option.

Anthropic is candid that evaluations are imperfect, and that both models voice suspicion of being tested more often than recent models.

Pricing and Availability

Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. Cache reads cost $0.20 per million tokens. A fast mode generates text up to 2.5 times faster at $16 and $80. Fable 5.2 is available on Pro, Max, Ultra, Team, and Enterprise plans, in Claude Code, on the Claude Platform, and on AWS, Google Cloud, and Microsoft. Mythos 5.2 is limited to verified organizations.

Which Model Should You Use?

Anthropic recommends starting with Opus 5.5 for most coding and knowledge work, then moving up to Fable 5.2 for the most ambitious projects or down to Sonnet 5.5 or Haiku 5.5 for lighter tasks. Choose Mythos 5.2 only if your organization does approved security or life sciences work. Test on a small real task first and measure cost per finished job, not just price per token.

Limits to Keep in Mind

Most benchmark numbers come from Anthropic, and some tests were run internally. Fable’s safeguards will block some legitimate requests. Long agent runs use many tokens and can be costly. Human review is still essential for important decisions.

Conclusion

Claude Fable 5.2 and Mythos 5.2 show how quickly AI is moving from answering questions to finishing projects. They share one engine, but different safeguards and access rules set them apart. The best results come from clear goals, sensible effort settings, human review, and honest testing.

Start small, measure what happens, and build from there.

Official Sources

FAQs

1. What Are Claude Fable 5.2 and Claude Mythos 5.2?

Claude Fable 5.2 and Claude Mythos 5.2 are advanced AI models introduced by Anthropic on November 9, 2026. They are designed for demanding reasoning, agentic coding, scientific research, cybersecurity, and complex engineering tasks. Both share the same underlying model capabilities but have different safety controls and access policies.

2. What Is the Main Difference Between Claude Fable 5.2 and Mythos 5.2?

The primary difference is their safety configuration and availability. Claude Fable 5.2 includes additional cybersecurity and biology safeguards and is available to general users. Claude Mythos 5.2 provides access to the same underlying capabilities without certain safeguards and is restricted to verified organisations working in areas such as cybersecurity and biomedical research.

3. Who Developed Claude Fable 5.2 and Mythos 5.2?

Both models were developed by Anthropic. The company focuses on developing advanced AI systems, improving model capabilities, and researching AI safety and responsible deployment.

4. How Do Claude Fable 5.2 and Mythos 5.2 Work?

These models process user instructions, analyse context, generate responses, and perform complex tasks using advanced reasoning capabilities. They can also work with tools, execute code, coordinate multiple AI agents, and handle extended workflows. Their adaptive thinking capabilities help them allocate reasoning effort to challenging tasks.

5. What Are the Key Features of Claude Fable 5.2?

Key features include advanced reasoning, agentic coding, long-horizon task execution, scientific research assistance, visual understanding, hardware engineering support, and multi-agent coordination. It also supports a large context window and extended output generation for complex projects.

6. What Is the Context Window of Claude Fable 5.2 and Mythos 5.2?

Both models support a context window of up to 1 million tokens and a maximum output of 128,000 tokens. This allows them to process extensive documents, large codebases, research materials, and complex instructions within a single workflow.

7. How Are Claude Fable 5.2 and Mythos 5.2 Different From Previous Models?

Compared with their 5.1 versions, the 5.2 models demonstrate improvements in coding, reasoning, scientific research, and visual understanding. Anthropic also reports faster output generation and improved efficiency. Their performance improvements are particularly relevant to complex, multi-step technical tasks.

8. Can Claude Fable 5.2 Perform Complex Coding Tasks?

Yes. Claude Fable 5.2 is designed to handle advanced software engineering tasks, including debugging, code generation, application development, testing, and large-scale code migrations. It can also coordinate multiple agents to work on different parts of a software project.

9. How Does Claude Mythos 5.2 Support Cybersecurity?

Claude Mythos 5.2 is designed to support advanced cybersecurity engineering. It can identify software vulnerabilities, analyse code, reproduce security issues, and help developers investigate and fix weaknesses. Its cybersecurity capabilities are available through Anthropic's verification programmes and eligible enterprise security offerings.

10. Can Claude Fable 5.2 Help With Scientific Research?

Yes. Claude Fable 5.2 can assist with scientific literature analysis, mathematical reasoning, research planning, simulations, and complex technical investigations. Anthropic reports that the model helped researchers make progress on challenging theoretical physics problems and mathematical formalisation.

11. How Does Claude Fable 5.2 Handle Mathematical Problems?

Claude Fable 5.2 can analyse mathematical expressions, develop proof strategies, write computational code, and assist with formal mathematical verification. Anthropic reports that a team of approximately 1,000 Fable 5.2 agents formalised a proof of Fermat's Last Theorem in the Lean proof assistant in 30 hours.

12. Can Claude Fable 5.2 Work With Multiple AI Agents?

Yes. Claude Fable 5.2 can coordinate multiple AI agents to divide complex projects into smaller tasks. Individual agents can perform research, write code, test results, and communicate progress. This approach can help manage projects that would be difficult for a single model to complete efficiently.

13. How Does Claude Fable 5.2 Support Hardware Engineering?

Claude Fable 5.2 can assist with electronic circuit design, component selection, PCB routing, computer-aided design, and embedded software development. Anthropic reports that it designed three working electronic devices, including a voice-assistant hub, a USB speakerphone, and a handheld virtual pet.

14. Can Claude Fable 5.2 Be Used in Robotics?

Yes. Claude Fable 5.2 has demonstrated improved capabilities in robotic task planning and control. Anthropic reports that it completed simulated and physical robot tasks involving object manipulation, puzzle solving, and other interactions. These capabilities could support future developments in autonomous robotics.

15. What Are the Safety Features of Claude Fable 5.2?

Claude Fable 5.2 includes safeguards designed to restrict potentially harmful cybersecurity and biological research activities. Anthropic also conducted pre-release alignment testing, behavioural evaluations, and external safety assessments to examine potential risks associated with the model.

16. Why Is Claude Mythos 5.2 Restricted to Verified Organisations?

Claude Mythos 5.2 provides access to advanced capabilities in sensitive areas such as cybersecurity and biology. These capabilities could be misused if made widely available without appropriate controls. Anthropic therefore restricts access through programmes that evaluate organisations and their intended applications.

17. How Much Do Claude Fable 5.2 and Mythos 5.2 Cost?

According to Anthropic's published API pricing, both models cost $8 per million input tokens and $40 per million output tokens. An optional fast mode is available at higher pricing: $16 per million input tokens and $80 per million output tokens. Actual costs depend on usage and applicable platform pricing.

18. Where Can Developers Access Claude Fable 5.2?

Developers can access Claude Fable 5.2 through the Claude API and supported cloud platforms, including Amazon Web Services and Google Cloud. Its API model identifier is claude-fable-5-2. Claude Mythos 5.2 requires eligibility through Anthropic's verification programmes.

19. What Industries Can Benefit From Claude Fable 5.2 and Mythos 5.2?

Potential applications include software development, cybersecurity, scientific research, mathematics, robotics, hardware engineering, financial research, and enterprise automation. Organisations can use these models to investigate complex problems, improve technical workflows, and support specialised research activities.

20. What Is the Future Potential of Claude Fable 5.2 and Mythos 5.2?

These models demonstrate how advanced AI could move beyond conversational assistance towards completing complex technical projects. Their applications in agentic coding, scientific discovery, cybersecurity, robotics, and hardware engineering could influence how organisations approach research, development, and automation. Their long-term impact will depend on continued improvements in reliability, safety, and responsible deployment.