After Fable 5.2’s price cut: keep Opus for daily work, use Fable for harder jobs, and note the extra usage window

The scene

After Claude Fable 5.2 launched on 9 Nov 2026, you want to know whether everyday coding still belongs on the strongest model, and how to use any extra Fable usage on a paid plan before it expires.

Tools involved

What they did

1. Follow the official split: start everyday coding and knowledge work on Opus 5.5; move up to Fable 5.2 for the hardest, open-ended technical jobs; send routing, file reads, and high-volume, well-specified subtasks to Haiku 5.5 or Sonnet 5.5. Fable can also act as an advisor that a smaller model consults, or as an orchestrator that hands simpler work to smaller models.
2. Compare the official API list prices checked on 10 Nov 2026 (USD per million tokens). Fable 5.2 is $8 input / $40 output, with $10 for a 5-minute cache write, $16 for a 1-hour cache write, and $0.20 for a cache hit. Fable 5.1 is $10 / $50 (cache hit $0.25). Opus 5.5 is $4 / $20 (cache hit $0.20). Sonnet 5.5 is $2 / $10 (cache hit $0.10). Haiku 5.5 is $0.10 / $0.50 when the prompt is at most 100,000 tokens (cache hit $0.01), and $0.50 / $2.50 above that. Batch is half those rates — for example $4 / $20 on Fable 5.2 and $2 / $10 on Opus 5.5.
3. Fable 5.2 also has an optional fast mode, listed at $16 input / $80 output. Fast mode for Opus 5.5 is $8 / $40. The pricing page says fast mode cannot be used with the Batch API. The launch post says Fable 5.2 generates outputs up to 30% faster than Fable 5.1; that is Anthropic’s wording, not a speed test run here.
4. On the API, the official prompting guide says to set effort to high (the API default) first, then try medium and xhigh on your own tasks. It says the step from medium to high raised scores on most of the coding and knowledge-work evaluations Anthropic tracks, at a higher token and time cost, and that gains above high are uneven. Use low for well-specified, high-volume subagents. At high and above, the model thinks before almost every reply, including a greeting, so drop effort on easy turns rather than adding a line that tells it not to think — in Anthropic’s testing, that line did not stop it thinking at medium and above.
5. If you are on a paid Claude plan, read the Help Center article “Claude Fable 5.2 launch offers” (the page showed it was updated today). Pro plans and standard Team seats: from launch on 9 Nov 2026 through 16 Nov 2026 at 11:59 pm Pacific Time, all Fable models are included for up to 50% of weekly usage limits, plus one Fable limit reset that expires at the same time. Max, Ultra, premium Team seats, and premium seats on legacy seat-based Enterprise plans: one to three Fable resets based on how long you have paid (one if under six months, two if six to 12 months, three if more than 12 months). Those resets expire on 8 Mar 2027 at 11:59 pm Pacific Time. Resets go only to accounts with paid billing in place before 9 Nov 2026 at 10:00 am Pacific Time; on Team and Enterprise, the organization’s billing must have been in place by then, and the seat holder must have joined by then. The Free plan, standard seats on legacy Enterprise, self-serve Enterprise, usage-based Enterprise, and HIPAA-ready Enterprise plans are not included.
6. Officially, a Fable reset does not raise your limits. It gives back weekly usage you have already spent on Fable models, and if your account has a five-hour window it refills that whole window, including usage spent on other models. Use one in Settings > Usage under Limit resets by choosing Reset for free. You can use it any time before it expires; once used, it cannot be undone. Your default model does not change — pick a Fable model yourself. The article also warns that Fable models use your limits faster than other Claude models.

Copyable prompt

Here is the kind of work I do this week: [describe the tasks]. Using only the official Claude prices and plan rules I paste below, recommend which model and effort level to use for each kind of task, and whether any time-limited Fable usage on my plan expires soon. If I did not paste a price or a date, say so and do not invent one.

Rewritten here for you to try; not a verbatim excerpt from elsewhere. What it asks：Give the model your task types plus the official prices and offer rules you pasted, and ask it which model and effort to use, and whether any extra Fable usage will expire soon.

Things to watch

The prices and offer rules above come from the official pricing page, the Fable 5.2 announcement, the Fable 5.2 prompting guide, and the Help Center offer article, all checked on 10 Nov 2026. They are standard API list prices per million tokens, before tax, and may not match your bill. Cache, batch, fast mode, region, and plan tier all change what you pay. Anthropic’s statement that Fable 5.2 at high effort scored about level with Opus 5.5 at max effort for roughly half the cost per task refers to its FrontierCode 2.0 evaluation, not a guarantee on your project. Offer dates are in Pacific Time. If an official page changes, use the new page.