Claude Fable 5.2 is the flagship model Anthropic released on November 9, 2026, built for the most demanding reasoning and long-running agent tasks. It is open to Pro, Max, Ultra, Team, and Enterprise users; the free plan does not include it. Look at the price and the entry point first, then decide whether it is worth upgrading for.
Work out the price first, then decide whether to use it
| Item | Fable 5.2 | Fable 5.1 |
|---|---|---|
| Input per million tokens | $8 | $10 |
| Output per million tokens | $40 | $50 |
| Change | 20% cheaper | — |
The context window is 1 million tokens and the maximum output is 128,000 tokens, fitting long documents and multi-step tasks. On long tasks, output tokens are often the biggest cost, so check the output price before estimating the total — don't look only at the input price. The API model ID is claude-fable-5-2; enter that ID directly when calling the API. On the subscription side it is even simpler: just pick Fable 5.2 in the model selector, with no other settings to change.
You can choose the model by task: everyday conversation and light writing don't need a flagship; save Fable 5.2 for deep reasoning, long chains, and multi-step agent work. Pro users spend paid quota when using Fable, but there is a launch-week discount running through 11:59 PM Pacific Time on November 16, 2026; after that, normal quota applies. Claude Code users need to upgrade to version 2.1.321 or higher and switch with the /model command.
Mythos 5.2 is easy to confuse with it: it shares the same underlying model as Fable 5.2 with a different safety configuration, and is only open to institutions that have passed the cybersecurity and life sciences verification program. Ordinary users can ignore it and just pick Fable 5.2. Free users who want to try the flagship should first weigh the cost of upgrading, rather than looking for a Fable 5.2 entry point inside the free plan.
