- Should you use Claude Fable 5.2 at all?
- Is Fable at High effort cheaper than Opus at Max?
- How do you use Claude Fable 5.2 step by step?
- Which prompts work best with Claude Fable 5.2?
- What's the difference between Fable 5.2 and Mythos 5.2?
- Which tools work with Claude Fable 5.2?
- What's my verdict on Claude Fable 5.2?
- Frequently asked questions
Use Claude Fable 5.2 for long, open-ended, technical work, and start everything else on Opus 5.5, which is Anthropic's own advice. Effort defaults to Medium in the Claude apps and Claude Code, so move it to High for hard jobs. At High, Anthropic reports it scored about level with Opus 5.5 at Max effort for roughly half the cost per task.
Use Claude Fable 5.2 for long, open-ended, technical work, and set it to High effort when you do. For almost everything else, start on Claude Opus 5.5, which is exactly what Anthropic's own launch post tells most people to do.
That's an odd thing for a company to say about the model it's just launched, and it's the most useful line in the whole announcement. So this guide starts with whether you need Fable at all, and only then gets to how to use it.
Should you use Claude Fable 5.2 at all?
Anthropic's advice is to start with Opus 5.5 for most coding and knowledge tasks, then go up to Fable or Mythos, or down to Sonnet 5.5 or Haiku 5.5, as the work needs. Here's how I'd apply that in a small business:
| The job | Where I'd start | Why |
|---|---|---|
| Long research that ends in a decision | Fable 5.2 at High | It's open-ended and the cost of a wrong answer is high |
| Turning a messy spreadsheet into a report | Fable 5.2 at High | There are many steps and plenty of ways for the data to mislead |
| A multi-step website audit | Fable 5.2 at High | It has to hold a long chain of checks together |
| Planning a migration | Fable 5.2 at High | Ambiguous, with dependencies that bite later |
| Everyday drafting and analysis | Opus 5.5 | Anthropic calls it its strongest daily driver |
| Quick rewrites and routine tidying | Sonnet 5.5 or Haiku 5.5 | The work doesn't need the extra thinking |
Who should use it: anyone with a hard, messy job they'd otherwise hand to a specialist or put off for a month. Who shouldn't: anyone who mainly wants emails, social posts and quick summaries. Opus 5.5 or Sonnet 5.5 will handle that work, and that's where Anthropic points most people. If you're unsure which of the everyday models suits you, my comparison of Opus and Sonnet will help.
Is Fable at High effort cheaper than Opus at Max?
On Anthropic's own numbers, yes, and this is the part most people will miss. The default effort in the Claude apps and Claude Code is Medium. At High effort, Fable 5.2 scored about level with Opus 5.5 at Max effort on FrontierCode 2.0, for roughly half the cost per task.
| Setting | FrontierCode 2.0 score | Cost per task |
|---|---|---|
| Opus 5.5 at Max | 45.5% | Baseline |
| Fable 5.2 at High | 45.3% | Roughly half of Opus at Max |
| Fable 5.2 at Xhigh | 49.1% | Not given as a comparison |
So if you're a heavy user running Opus at Max, the cheaper move is Fable at High, not more Opus. Anthropic also reports 51.4% for Fable 5.2 on AutomationBench against 42.5% for Opus 5.5. All of these are Anthropic's own figures, so treat them as a reason to test it on your work, not as proof.
If you're paying through the API, here's the price list as of Nov 2026, in US dollars:
| Mode | Input, per million tokens | Output, per million tokens |
|---|---|---|
| Standard | $8 | $40 |
| Fast mode, up to 2.5 times faster generation | $16 | $80 |
Fast mode doubles the bill, so I'd save it for jobs where someone is waiting on the answer. It's available on Claude Code, the Claude Platform, Claude Models on Google Cloud and Microsoft Foundry.
How do you use Claude Fable 5.2 step by step?
- Check the job against the table above. If it isn't hard and open-ended, start on Opus 5.5 instead.
- Choose Fable 5.2 where you work. That's the Claude apps, Claude Code or Cowork. Developers use the model name claude-fable-5-2 on the Claude Platform, and it's also on Amazon Web Services, Google Cloud and Microsoft Azure.
- Move effort from Medium to High. Medium is the default, and High is where the comparison above applies.
- Give it the whole problem in one go. Include the goal, the material and what a good result looks like.
- Ask for assumptions first. Correct them before it does the long part of the work.
- Check the output against your own knowledge. Open the numbers and the sources yourself before you act on any of it.
- Hand the routine follow-ups down. Use a smaller model for the rewrites and tidying.
One timing note: this launch week, as of Nov 2026, subscribers on eligible paid plans get extra Fable usage, with one to three Fable resets depending on subscription length. Fable is also included for up to 50% of weekly usage limits for Pro plans and standard Team seats. That makes this a good week to try it on your hardest job.
Which prompts work best with Claude Fable 5.2?
Fable earns its keep on briefs with a decision at the end. These four are mine, so swap in your details at each bracket.
For long research that has to end in a decision
Copy this prompt
I need to decide [decision] by [date]. The topic is [topic].
Research this as if you were advising me personally. Tell me what would change your recommendation.
Separate what you're confident about from what you're inferring.
End with one recommendation, the two strongest arguments against it, and the first step I should take on Monday.
For an ambiguous job with many moving parts
Copy this prompt
I want to [goal], and I haven't worked out how yet. Here's what I know: [context].
Before you plan anything, list every assumption you're making and number them. Then stop and wait for me to correct them.
After I reply, give me a plan in stages, with what could go wrong at each stage and how I'd notice.Work with meWant AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
For reviewing a plan made by a cheaper model
Fable 5.2 can work as an Advisor, where Haiku, Sonnet or Opus models consult it on harder problems, or as an Orchestrator, delegating simpler tasks to smaller models. You can get the same idea by hand with this:
Copy this prompt
Here's a plan a colleague drafted for [project]: [paste the plan].
You're reviewing it, not rewriting it. Name the three risks most likely to sink it, in order.
For each one, tell me what evidence would show it's happening and the smallest change that would reduce it.
For cleaning up messy data
Copy this prompt
This is [file description], and it's a mess: [paste the data or describe the columns].
First tell me what's inconsistent, missing or duplicated, without changing anything.
Then give me the cleaned version, plus a list of every change you made and every row you weren't sure about.
Finally, write a one-page summary for [audience] that only uses figures you can trace back to the data.
What's the difference between Fable 5.2 and Mythos 5.2?
They're the same underlying model. Fable has strong cybersecurity and biology safeguards, so it's open to all users. Mythos access is by application, through the Cyber Verification Program or the Life Sciences Verification Program. If you don't work in those fields, Fable is the one you want. I've explained more about what Claude Mythos is separately.
Which tools work with Claude Fable 5.2?
- Claude Code for building and fixing, with fast mode available.
- Cowork as another place Fable 5.2 runs, alongside the Claude apps.
- Claude Opus 5.5 as the place to start for most jobs.
- Claude Sonnet 5.5 and Claude Haiku 5.5 for the routine follow-ups.
- Google Sheets as the home for the data you ask it to clean.
- Zapier for moving a finished result on to the next app.
What's my verdict on Claude Fable 5.2?
I'd use it, but not as my default, and after 23 years in marketing I'm wary of anyone who tells you the newest thing should do every job. Keep Opus 5.5 as your everyday model, and bring in Fable at High effort for the two or three hard jobs a month that would otherwise eat a day. If you're paying for Opus at Max, test the swap this week.
If you'd like help working out which jobs in your business are worth it, this is how I work with business owners.
Frequently asked questions
Is Claude Fable 5.2 faster than Fable 5.1?
Yes. Anthropic says outputs are up to 30% faster than Fable 5.1 and Mythos 5.1. Optional fast mode generates up to 2.5 times faster, at double the standard API price as of Nov 2026.
What effort setting should I use on Fable 5.2?
The default in the Claude apps and Claude Code is Medium. I'd use High for hard jobs, because that's where Anthropic reports it matching Opus 5.5 at Max for roughly half the cost per task.
Will Fable 5.2 refuse fewer requests than 5.1?
Anthropic reports about 55% fewer safeguard blocks than Fable 5.1 in early testing across claude.ai, Claude Code and Cowork.
What's the API model name for Fable 5.2?
It's claude-fable-5-2 on the Claude Platform. As of Nov 2026, the API price is $8 per million input tokens and $40 per million output tokens.
Does Anthropic keep my data when I use Fable 5.2?
Zero data retention customers can continue using it with zero retention before safety monitoring with EFS. Other customers get 30-day retention for safety monitoring.
