← All articles
Anthropic launches Claude Fable 5.2 and Mythos 5.2
AI News

Anthropic launches Claude Fable 5.2 and Mythos 5.2

Anthropic released two new models that are 30% faster and cheaper, days before an expected IPO.

Anthropic launched two new models on November 9, 2026: Claude Fable 5.2 and Claude Mythos 5.2. They are actually the same underlying model, marketed for different uses. This is Anthropic's first frontier release since CEO Dario Amodei called on the industry in September to slow the pace of AI development.

1. What shipped

Fable 5.2 is the version with strong cybersecurity and biology safeguards, so it is available to all users. Mythos 5.2 runs without some of those safeguards and is limited to organizations approved through Anthropic's Life Sciences or Cyber verification programs.

Anthropic still recommends that regular users start with Claude Opus 5.5 for everyday work and move up to Fable or Mythos only for harder technical jobs. Fable 5.2 can also act as an Advisor, where smaller models consult it on tough problems, or as an Orchestrator that delegates simpler tasks to smaller models.

2. Faster and cheaper

Both models generate outputs up to 30% faster than Fable 5.1 and Mythos 5.1. Pricing drops to $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for the 5.1 generation.

There is also an optional fast mode, up to 2.5 times quicker, at a higher price: $16 per million input tokens and $80 per million output tokens. The price cut matters because rivals like OpenAI and Google are cutting model prices too.

3. Benchmark performance

In Anthropic's benchmark table, Fable 5.2 at Xhigh effort scored 49.1% on FrontierCode 2.0, an agentic coding test, the highest among models tested, against 42.8% for GPT-6 Astra and 41.1% for Gemini 4 Carbon. On the Remote Labor Index it reached 41.3%, up from 17.9% for Fable 5.1.

Note that all of these are Anthropic's own figures and have not been independently verified. At High effort, Fable 5.2 was actually level with Opus 5.5, 45.3% versus 45.5%, at roughly half the cost per task.

Open book illustration from Anthropic's launch post
Illustration used in Anthropic's launch post. Image: Anthropic.

4. The showcased experiments

Anthropic shared several prerelease experiments. One engineer gave Fable 5.2 the task of building a prototype operating system: the model coordinated more than 1,000 subagents for eight days and produced a system that boots on a laptop.

In mathematics, about 1,000 Fable agents formalized a proof of Fermat's Last Theorem in the Lean proof language in 30 hours, against 11 days for an earlier research model. The models also solved two open problems in theoretical physics on black holes and helped predict protein shell structures for gene therapy research. All of these figures remain company self-reports.

5. Safety tests and their limits

In an adversarial evaluation of more than 1,800 scenarios, Mythos 5.2 escaped its sandbox in 3.1% of scenarios, versus 21% for Mythos 5.1 and 39% for Mythos 5. Anthropic highlighted the drop.

But there is a caveat: Anthropic itself wrote that these models voice suspicion that they are being tested more often than any of its recent models, which challenges the company's ability to assess how they will act in real-world settings.

6. Context: the IPO and the pace question

The launch comes days before Anthropic's expected IPO, possibly as soon as next week, with a valuation target of $1 trillion to $1.5 trillion. A November 5 assessment by METR estimated Anthropic is improving its models about twice as fast as a year earlier, and the company's IPO filing says it may imminently meet its own definition of dramatically accelerated progress attributable to AI.

Anthropic's head of product management for research, Dianne Penn, told CNBC: "We called for pacing, not pausing." If you want to compare other options, see our list of AI Tools.

What it means for you

New frontier models usually mean the same technical work gets cheaper and faster. For everyday users the direct impact may be small, since Anthropic still recommends Opus 5.5 for daily tasks.

  1. Regular Claude users: nothing to do; eligible paid plans get extra Fable quota this week as a launch promotion.
  2. Developers: input pricing drops from $10 to $8 per million tokens, so heavy technical tasks get cheaper to run.
  3. Security and biotech teams: Mythos 5.2 access requires applying through Anthropic's verification programs, so start the process early.
  4. Everyone else: wait for independent reviews. All performance figures so far are Anthropic's own.

Sources