Claude Fable 5.2: what's new, what it costs, and Fable 5.2 vs Opus 5.5

Short answer: Claude Fable 5.2 is Anthropic's most capable model, released on November 9, 2026, for the hardest and longest agent work. It beats Fable 5.1 by a wide margin on coding (69.9% against 56.3% on FrontierSWE v2), costs 20% less on the API ($8 and $40 per million tokens, down from $10 and $50), answers up to 30% faster and refuses about 55% less often. On Max and Ultra it's included for up to half your weekly limit; on Pro it's paid with usage credits, except during a launch week that runs until November 16. In Claude Code, type /model fable. Anthropic still recommends Opus 5.5 for everyday work.

Everything below is from Anthropic's announcement, its model documentation and the Claude Help Center, checked on November 10, 2026. (I build Morphy, which runs Claude Code next to other AIs; it comes up once near the end.)

What's new in Fable 5.2

Better at long, hard work. Anthropic built it for agent sessions that run for hours, research in many steps, and analysis carried through to a finished document. Its numbers jumped most on agentic coding, computer use and knowledge work (table below).
Cheaper. $8 per million input tokens and $40 per million output tokens on the API, against $10 and $50 for Fable 5.1. Cached input costs $0.20 per million, and the Batch API halves the price.
Faster. Answers come up to 30% faster than Fable 5.1. Fast mode makes it up to 2.5 times faster again, at $16 and $80 per million tokens.
Fewer refusals. In Anthropic's testing across claude.ai, Claude Code and Cowork, Fable 5.2 blocked about 55% fewer requests than Fable 5.1. When its cyber or biology safeguards do block one, the request falls back to Opus 4.8 or Opus 5.
The same shape. A 1-million-token context window, up to 128,000 tokens of output, knowledge through June 2026, and thinking that's always on, set by effort.

And Mythos 5.2? It's the same model without the cyber and biology safeguards, only for organizations Anthropic has verified, such as through its Cyber Verification Program, and Enterprise customers using Claude Security. Everyone else gets Fable 5.2.

Fable 5.2 vs Opus 5.5 vs Fable 5.1

Anthropic's own benchmarks, from the announcement:

Benchmark	Fable 5.2	Fable 5.1	Opus 5.5	GPT-6 Astra	Gemini 4 Carbon
FrontierSWE v2 (agentic coding)	69.9%	56.3%	62.3%	65.5%	61.2%
FrontierCode 2.0 (agentic coding)	49.1% (xhigh effort)	32.4%	45.5%	42.8%	41.1%
OSWorld 2.1 (computer use)	89.3%	83.5%	87.2%	79.5%	81.5%
Humanity's Last Exam, with tools	74.2%	65.6%	67.7%	57.2%	–

So Fable 5.2 leads every one of these, and the jump from Fable 5.1 is bigger than its lead over Opus 5.5. That lead costs twice as much per token as Opus 5.5 ($8 and $40 against $4 and $20).

Which to use: Anthropic's own advice is to start with Opus 5.5, "our strongest daily driver", and move to Fable 5.2 when Opus at a higher effort still falls short. In practice: Opus 5.5 (or Sonnet 5.5) for everyday coding; Fable 5.2 for root-cause hunts, outage debugging, architecture decisions and tasks you'd normally break into pieces.

Can you use Fable 5.2 on your plan?

Plan	Fable 5.2
Free	No
Pro, Team standard seats	Paid with usage credits. Launch week: included, up to half your weekly limit, until November 16, 11:59 pm PT
Max, Ultra, Team premium seats	Included, up to half your weekly limit, plus one to three free Fable resets
Enterprise	Off until an admin turns it on; from November 16, added wherever a Fable model is already on, unless an admin opts out
API key	$8 / $40 per million tokens

Two things to know about "up to half your weekly limit": it isn't extra usage (Fable draws from the same weekly limit as everything else, just faster), and past that half you either switch models or pay with usage credits.

Fable resets

For the launch, Anthropic gives eligible paid plans free Fable resets. A reset puts your Fable usage on the weekly limit back to zero; if your plan has a 5-hour limit, it refills that window too, for every model.

Max, Ultra and premium seats: one reset if you've subscribed for less than six months, two for six to twelve, three for longer. They expire on March 8, 2027.
Pro and standard Team seats: one reset, which expires with the launch week on November 16.
To use one: Settings → Usage on claude.ai, then Reset for free under Limit resets. You can use it any time before it expires, not only at a limit, and it can't be undone.

You needed a paid plan before November 9, 10:00 am PT to qualify.

How to use Fable 5.2 in Claude Code

Update Claude Code. Fable 5.2 needs version 2.1.321 or later: run claude update.
Pick it: /model fable in a session, or claude --model fable to start one. The fable alias now means Fable 5.2.
Choose the effort. Claude Code and the Claude apps run Fable 5.2 at medium by default; the API default is high. Anthropic says most of the gain is there at high, xhigh is for the hardest long-running work, and max for the rare task where time and tokens don't matter. Set it with /effort.

When Fable would bill to usage credits (on Pro, after the launch week), the /model picker shows Requires usage credits on the Fable row. Once you agree to continue with usage credits, Claude Code doesn't ask again.
Saved Fable 5.1? On Pro and Max, a saved claude-fable-5-1 moves to the fable alias the first time you run 2.1.321 or later. To stay on 5.1, use /model claude-fable-5-1.
Fast mode: /fast turns it on, while you're on Fable 5.2. On a plan it's always paid with usage credits, even with plan usage left, and switching mid-conversation re-reads the whole conversation at the fast price, so turn it on at the start.
Hand it outcomes, not steps. Anthropic's tips for Claude Code: describe the result you want, give it ambiguous problems, skip the reminders to test (it checks its own work), and give it bigger tasks than you would Opus.

In the Claude apps, pick Fable 5.2 in the model picker. Fable 5.1 and Fable 5 stay available.

For developers: what changed in the API

The model ID is claude-fable-5-2, on the Claude API, Amazon Bedrock (anthropic.claude-fable-5-2), Google Cloud and Microsoft Foundry. Four changes from Fable 5.1 can break code that builds its own requests (nothing changes if Claude Code or the Agent SDK builds them):

Editing earlier turns invalidates thinking blocks on every account, with a 400 error saying The block is bound to a different conversation. Keep system, tools and earlier messages unchanged and only append.
Only Fable 5.2 and Mythos 5.2 read Fable 5.2's thinking blocks, so switching a conversation to another model loses that reasoning.
The older computer_20251124 tool returns an error (except on Bedrock): use the computer_toolset_20260801 toolset.
With the advisor tool, a Fable 5.2 executor needs a Fable 5.2 or Mythos 5.2 advisor.

Forced tool use (tool_choice any or tool) and temperature, top_p or top_k return errors, thinking can't be turned off, and Fable 5.2 needs 30-day data retention.

In Morphy

In Morphy, each chat picks its own model from the menu under its text box, so one chat can run Fable on the hard problem while the others stay on Opus or Sonnet, or on another AI. Settings → Plan usage shows how much of your Claude plan is left and when it resets. See Models and effort.

Questions

When was Claude Fable 5.2 released? November 9, 2026, together with Mythos 5.2.

How much does Fable 5.2 cost? On the API, $8 per million input tokens and $40 per million output tokens, 20% less than Fable 5.1. On Max and Ultra it's included up to half your weekly limit; on Pro it's paid with usage credits after the launch week.

Is Fable 5.2 better than Opus 5.5? On every benchmark in Anthropic's announcement, yes, by about 4 to 8 points on coding. It also costs twice as much per token and thinks longer, so Anthropic recommends Opus 5.5 for daily work and Fable 5.2 for the hardest tasks.

How do I use Fable 5.2 in Claude Code? Run claude update (it needs 2.1.321 or later), then /model fable.

What's a Fable reset? A free launch bonus that sets your Fable usage on the weekly limit back to zero. Use it in Settings → Usage on claude.ai.

What's the difference between Fable 5.2 and Mythos 5.2? The same model; Mythos has no cyber and biology safeguards and is only for verified organizations.

References

Anthropic: introducing Claude Fable 5.2 and Claude Mythos 5.2, Claude Fable 5.2, what's new in Fable 5.2, models overview, plans and pricing
Claude Help Center: Fable models on your plan, Fable 5.2 launch offers
Claude Code: model configuration, fast mode