Claude Fable 5.2 Is Faster, Cheaper and Tops the Coding Charts

What happened

On November 9, Anthropic released Claude Fable 5.2, along with Claude Mythos 5.2, which is built on the same base model. Fable 5.2 is available to all Claude API users from day one, and it can also be called directly on AWS, Google Cloud and Microsoft Azure. Mythos 5.2 is limited to organizations approved through Anthropic's cybersecurity and life sciences verification programs.

According to Anthropic's launch page and developer docs, Fable 5.2 changes these things:

• Faster: Generation speed is up to 30% faster than Fable 5.1.
• Cheaper: $8 per million input tokens and $40 per million output tokens. Fable 5.1 cost $10 and $50, so both prices dropped by 20%.
• Top coding score: On FrontierCode 2.0, Fable 5.2 scored 49.1% at the Xhigh effort level. Anthropic says this is the highest score of any released model it has tested. In the same test, Fable 5.1 scored 32.4%, Opus 5.5 scored 45.5%, and GPT-6 Astra scored 42.8%.
• New fast mode: Generation speed rises to up to 2.5 times faster, at $16 per million input tokens and $80 per million output tokens, which is exactly twice the standard price. It is available in Claude Code, the Claude API, Google Cloud and Microsoft Foundry.
• Fewer false refusals: Harmless requests that were wrongly blocked dropped by about 55% compared with Fable 5.1.

Why it matters

A top model upgrade usually means more capability and a higher price. This time the model runs faster, costs less, and still takes the top coding score.

But don't treat it as the new default. Anthropic recommends using Opus 5.5 first for most coding and knowledge work. Switch to Fable 5.2 when Opus, even at a high effort level, can't solve the problem. Fable is strongest on long, open-ended, hard technical work.

One detail is easy to miss: 49.1% is the score at the top effort level. Anthropic also reports that Fable 5.2 scores 45.3% at the High level. That is close to Opus 5.5's best result, and each task costs about half as much. So you don't always need to run at the highest setting.

Getting started

The API model name is claude-fable-5-2. It has a 1M token context window and can output up to 128K tokens in one response. Thinking effort has five levels: low, medium, high, xhigh and max. The API defaults to high, while the Claude app and Claude Code default to medium. Pro users and standard Team seats can spend up to half of their weekly limit on Fable.

Not sure which model to use? Use this table:

Notes

• Benchmarks are self-reported: All the numbers above come from Anthropic's launch page. Independent third-party tests are still coming out.
• Fast mode costs twice as much: Turn it on only when you really need the speed.
• The API has breaking changes: For example, the thinking budget parameter and some tool_choice values now return errors. Read Anthropic's migration guide before you upgrade.
• Data retention: Fable 5.2 requires 30 days of data retention and does not support zero data retention by default. Enterprise users should check this.
• Mythos 5.2 is not available on standard accounts: You must first pass Anthropic's verification program.

References

• Anthropic: Introducing Claude Fable 5.2 and Claude Mythos 5.2
• Claude developer docs: What's new in Claude Fable 5.2