Anthropic ships Claude Fable 5.2 with 55% fewer blocks

Anthropic's top model now says yes more often.

Claude Fable 5.2 went to every Claude user and the major clouds on November 9, and Anthropic says its safeguards block far fewer requests than the last version did. That is the change you will notice first.

Fable 5.2 is open to everyone, Mythos 5.2 is not

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9. The company says they are the same underlying model.

Fable carries stronger cybersecurity and biology safeguards and is open to all users. Mythos goes only to organizations accepted into Anthropic's Cyber Verification Program or Life Sciences Verification Program.

Anthropic says Fable 5.2 is available on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure.

The API price, per million tokens:

$8 input, $40 output.

Fast mode, which Anthropic says generates tokens up to 2.5x faster, costs $16 input and $80 output. Even without it, Anthropic says outputs are up to 30% faster than Fable 5.1.

Fewer refusals is the upgrade that changes your day

Yes, another frontier model with another benchmark table. This one matters because it says no less often.

Anthropic says that in early testing Fable 5.2 blocks about 55% fewer requests overall than Fable 5.1, about 60% fewer in cybersecurity and about 40% fewer in biology. When it does block, the request is answered by an older model instead: Opus 4.8 for cybersecurity, Opus 5 for biology, according to the announcement.

Every block is a task you redo somewhere else. For security teams and life sciences work, that number beats any leaderboard.

The benchmarks are strong too, by Anthropic's count. It reports Fable 5.2 at 69.9% on FrontierSWE v2, against 56.3% for Fable 5.1 and 65.5% for OpenAI's GPT-6 Astra. Those are Anthropic's own figures.

The independent Artificial Analysis listing scores Fable 5.2 at its Max setting at 65 on its Intelligence Index and ranks it first of 74 models, with a 1 million token context window.

Anthropic itself says most work starts on Opus 5.5

Read the fine print before you move everything over. Anthropic's own post recommends starting with Claude Opus 5.5 for most coding and knowledge tasks, and moving up to Fable or Mythos for harder work.

So Fable 5.2 is for the hard jobs, and it is priced like it.

The details builders will want, from the announcement:

• The API model ID is claude-fable-5-2.
• Medium effort is the default in the Claude apps and Claude Code.
• Fast mode runs in Claude Code, the Claude Platform, Claude Models on Google Cloud and Microsoft Foundry.
• Pro and standard Team seats can spend up to 50% of their weekly usage limits on Fable.
• Eligible customers can use zero data retention. Other customers have 30 day retention for safety monitoring.

Anthropic also says its anti-distillation protections cut the data distillers harvest by 90% to 99%, a defense it first described in October.

Watch your refusal rate, not the leaderboard

The 55% figure comes from Anthropic's early testing. Your own logs are the real test.

Mythos is the other thing to watch. Anthropic says Mythos 5.2 found 10.2% more vulnerabilities than Opus 5.5 on its internal CVE Rediscovery evaluation of 1,829 CVEs. Claude Security, for Claude Enterprise customers, scans codebases with the Claude model family, Mythos 5.2 included. Outside testers included the UK AI Security Institute and METR, the company says.

Count your blocked requests this week. Then decide which jobs earn the $40.