Anthropic unveils Claude Fable 5.2 and Mythos 5.2 for harder technical work
Anthropic
The company that develops Claude.
token
A small piece of text processed by an AI.
verification program
A program that checks organizations before access.
What happened
Anthropic, the company behind Claude, announced Claude Fable 5.2 and Claude Mythos 5.2 on November 9. The two products use the same underlying model. They differ in access and safety settings. Fable 5.2 includes stronger safeguards for cybersecurity and biology requests. It is available broadly. Mythos 5.2 is offered through verification programs for approved organizations. It does not include some of Fable’s safeguards.
The background
Anthropic presents both models as tools for very difficult technical challenges. The company recommends Claude Opus 5.5 for most everyday coding and knowledge work. It suggests moving up to Fable or Mythos when a task becomes more open-ended or demanding. That makes the new models specialized products, rather than simple upgrades for every user.
What improved
Anthropic says the models are better at coding, reasoning, and vision. Their outputs can be up to 30% faster than those from the 5.1 versions. The listed price is $8 per million input tokens and $40 per million output tokens. Fast mode costs $16 and $80. A token is a small piece of text processed by an AI system.
The company also described several prerelease tests. Fable 5.2 helped with software migrations and designed three electronic devices. It made progress on open problems in black-hole physics. A large group of agents also formalized Fermat’s Last Theorem in a computer-checkable form. These examples are meant to show long, multi-step work, not only short answers.
Why the launch matters
Anthropic says Fable can serve as an advisor for hard tasks. It can also act as an orchestrator, giving simpler tasks to smaller models. This positions Claude as part of a larger workflow. The model may plan, write code, inspect results, and coordinate other systems.
The launch also makes access control part of the product design. Anthropic is trying to offer stronger abilities widely through Fable. It is reserving some higher-risk capabilities for organizations that pass extra checks through Mythos. The balance between usefulness and control is therefore central to the announcement.
What is confirmed
Fable 5.2 is available across Claude’s platforms. It is also offered through Amazon Web Services, Google Cloud, and Microsoft Azure. Mythos 5.2 is available through Anthropic’s Cyber Verification Program and Life Sciences Verification Program. Anthropic says its prerelease safety tests matched or beat recent models on most alignment measures.
What remains unknown
Most performance numbers come from Anthropic’s own tests. Some evidence comes from early testers who worked with the company. Independent researchers have not yet shown whether all the results repeat. It is also unclear how consistently the models will handle messy, real-world technical projects.
What to watch next
The next useful evidence will include independent benchmarks and customer results. Observers will also watch how widely Mythos access expands. Another key question is whether the safeguards can reduce needless blocks without enabling unsafe work. The launch is promising, but its broad impact still depends on use outside Anthropic’s testing environment.
Anthropic makes two stronger Claude models
📰 Full story: Anthropic unveils Claude Fable 5.2 and Mythos 5.2 for harder technical work
Anthropic, the company behind Claude, announced two models for difficult technical work.
Anthropic
The company that makes Claude.
token
A small piece of text read by an AI.
Mythos 5.2
The model for organizations that pass extra checks.
💡 The gist
- Anthropic announced Fable 5.2 and Mythos 5.2.
- They are better at coding, reasoning, and viewing charts.
- Fable is broadly available; Mythos needs extra approval.
Anthropic built both models from the same underlying model. Their safety settings are different. Fable includes extra protections for risky cybersecurity and biology requests. Mythos is for organizations that pass Anthropic’s checks.
This split has a reason. Stronger AI can help with important work. It can also create more risks in sensitive fields. Anthropic says Fable’s protections make it suitable for wider use. Mythos is limited to approved research and security programs.
Anthropic says the new models produce answers up to 30% faster. Their prices are also lower than the earlier versions. The basic price is $8 for one million input tokens. It is $40 for one million output tokens. A token is a small piece of text that an AI processes. A faster mode costs more.
The company tested the models on difficult projects. They helped move software from one system to another. They designed electronic devices. They worked on physics questions. They also helped turn a famous math proof into steps a computer can check.
These tasks are important because they require many steps. The AI must plan, write, inspect, and revise work. Anthropic says Fable can also guide smaller models. This could make it useful as a team leader for AI tasks.
Still, the evidence has limits. Anthropic produced most of the test results. Early testers also shared positive comments. Independent researchers have not confirmed every claim. We do not yet know how reliable these models will be in ordinary workplaces.
The next evidence should come from independent tests and real customers. People will also watch Mythos access and safety results. The main question is simple. Can these models stay useful without becoming too risky?
Claude has two new helpers
📰 Full story: Anthropic unveils Claude Fable 5.2 and Mythos 5.2 for harder technical work
Anthropic, the company that makes Claude, made two new AI helpers.
Anthropic
The company that makes Claude.
Fable 5.2
A Claude helper that many people can use.
Mythos 5.2
A Claude helper for checked teams.
Anthropic made two new helpers.
They are called Claude Fable 5.2 and Claude Mythos 5.2.
Both helpers have the same AI inside.
They can write computer instructions.
They can think through hard questions.
They can look at pictures and charts.
Fable has more safety rules.
Many people can use Fable.
Mythos is for teams that pass Anthropic’s checks.
Anthropic says they work faster than before.
The company tested them on difficult jobs.
Tests cannot show every real job.
People still need to learn how they work outside tests.