In short: Anthropic released Claude Fable 5.2 on November 9, 2026. The company positions it as the best choice for the most ambitious, open-ended technical work, and still recommends Opus 5.5 for everyday coding and knowledge tasks. So opening Fable for every job doesn’t make sense.
What we know
- Release: November 9, 2026.
- Context: 1M token context window, 128K token output.
- Thinking: adaptive thinking is always on; default effort is
high. - API model ID:
claude-fable-5-2.
Compared with Opus 5.5
| Benchmark | Fable 5.2 | Opus 5.5 |
|---|---|---|
| FrontierSWE v2 (coding) | 69.9% | 62.3% |
| OSWorld 2.1 (computer use, offline set, partial) | 89.3% | 87.2% |
Numbers are from Anthropic’s own announcement.
Fable vs Mythos 5.2
They are the same model. Fable ships with strong cyber and bio safeguards and is open to everyone. Mythos 5.2 goes to organizations accepted into Anthropic’s verification programs.
Plans and API
- Pro and standard Team seats: Fable is included for up to 50% of your weekly usage limit.
- Launch week: eligible paid plans get 1–3 Fable resets depending on subscription length.
- API price: $8 input / $40 output per million tokens; the Batch API is 50% off.
- Fast mode: $16 input / $80 output, up to 2.5x faster.
- Data retention: requires 30-day retention, so it isn’t available under zero data retention.
When to pick it
- Quick questions, one-file fixes: Sonnet.
- Everyday coding and code review: Opus 5.5.
- The hardest, open-ended, long technical work: Fable 5.2.
In Claude Code, type /model to pick Fable and use /effort to set the thinking level. Keeping effort low on small tasks saves your limit.
Free PDF guide
The 3-page guide covers which model to use for which job and how to protect your limit (in Turkish).
Source: Anthropic announcement and Fable 5.2 docs. Accurate as of November 14, 2026.
