Claude Fable 5.2 is out, 20% cheaper. Pro plans include it through November 16

What Anthropic released

On November 9, 2026, Anthropic released Claude Fable 5.2 and Claude Mythos 5.2. The two are “the same underlying model.” Fable “has strong cybersecurity and biology safeguards to guard against misuse, which allows us to make it available to all users.” Mythos goes only to organizations in Anthropic’s Cyber and Life Sciences Verification Programs.

Anthropic says the new models “communicate more clearly than past Fable and Mythos models” and “generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1.” Its Fable page says Fable 5.2 “is available to Pro, Max, Ultra, Team, and Enterprise users,” and to developers as claude-fable-5-2. The model page gives it a 1 million token context window, 128,000 output tokens, thinking that can’t be turned off, and a knowledge cutoff of June 2026.

It isn’t meant to be the model you use all day. “For most coding and knowledge tasks, we recommend that users start with Claude Opus 5.5, our strongest daily driver,” the announcement says, moving “upward to Fable or Mythos” when the work calls for it. Dianne Penn, Anthropic’s head of product management for research, told CNBC that Fable 5.2 is mainly for “power users or power use cases.”

What it costs

Model	API, in / out per million tokens	Fast mode, in / out	On Claude’s plans
Claude Fable 5.2	$8 / $40	$16 / $80	Max and Ultra: up to half the weekly limit. Pro: usage credits, after this week’s offer
Claude Fable 5.1	$10 / $50	None	The same as Fable 5.2
Claude Opus 5.5	$4 / $20	$8 / $40	Every paid plan, within its limits

The prices are from Anthropic’s API pricing and Claude Code’s fast mode guide. A cache read on Fable 5.2 costs $0.20 per million tokens, batch processing halves the rate to $4 and $20, and Mythos 5.2 costs the same as Fable 5.2. Fast mode is “up to 2.5x faster” at twice the price.

Fable 5.2 still costs twice what Opus 5.5 does per token. Anthropic’s case is that it needs fewer of them: in early testing on its FrontierCode 2.0 evaluation, “Fable 5.2 at High effort scored about level with Opus 5.5 at Max effort, for roughly half the cost per task.”

On a subscription, Claude’s pricing lists Fable as “Usage credits” on Pro and “50% of weekly limits” on Max and Ultra, and on Team for premium seats only. The half isn’t extra. Anthropic’s help article says “your use of other models draws from the same usage limits and you can never use more than your weekly limit.” Usage credits “are billed at standard API rates,” so on Pro, outside the offer below, Fable 5.2 costs $8 and $40.

What each plan gets this week

Anthropic is marking the launch with two offers, and which one you get depends on your plan.

Plan or seat	Offer	Ends
Pro, and standard seats on Team	Every Fable model included, up to 50% of weekly limits, plus one Fable limit reset	November 16, 2026, 11:59 pm PT
Max, Ultra, and premium seats on Team and legacy seat-based Enterprise	One to three Fable limit resets	March 8, 2027, 11:59 pm PT
Free, standard seats on legacy seat-based Enterprise, and self-serve or usage-based Enterprise	None

Both offers cover “Fable 5.2, Fable 5.1, and Fable 5.” The details that decide what you get:

• How many resets. On Max, Ultra and premium seats: “Under six months gets one reset, six to 12 months two, and more than 12 months three,” counted from your first paid bill. Canceling and coming back doesn’t restart the count.
• The cutoff. “Resets go only to accounts with paid billing in place before November 9, 2026 at 10:00 am PT.” Subscribing now for the resets doesn’t work. The article ties that cutoff to resets only; it sets no such condition on Pro’s included week.
• What a reset does. On the weekly limit, it “gives you back the usage you’ve spent on Fable models. Usage you’ve spent on other Claude models stays where it is.” On the five-hour limit, if your account has one, it “refills the whole five-hour window, including usage spent on other models.” It “doesn’t raise your limits.”
• How to use one. Go to Settings, then Usage, and under Limit resets select “Reset for free.” You can’t undo it, and “If you cancel or downgrade before using a reset, it’s no longer available.”

If you’re on Pro, three things are easy to miss:

1. Nothing switches for you. “Your default model doesn’t change. To use a Fable model, select it from the model picker.”
2. Fable drains the week faster. “Fable models use your usage limits faster than other Claude models,” Anthropic warns, without giving a ratio.
3. It can start costing money without a new decision. “After you use that 50%, or once the week ends, Fable usage is billed to usage credits if you have them turned on.” If you don’t want that, turn usage credits off or set a spend limit before you start.

How to use it in Claude Code

1. Update. “Fable 5.2 requires Claude Code v2.1.321 or later,” says the model guide. Run claude update.
2. Select it. Run /model fable, or start with claude --model fable. “No Fable model is the account-type default on any plan or provider,” so nothing changes until you choose it. The default on Claude’s plans is still Opus 5.5.
3. Check a saved Fable 5.1 pick. The changelog for 2.1.321 says it “Changed saved Fable 5.1 model picks to move to fable: on Pro and Max plans at once, on Team and Enterprise plans once your organization allows it; use /model claude-fable-5-1 to switch back.” Console accounts and third-party providers keep claude-fable-5-1.
4. Mind the effort level. Fable 5.2 starts at medium in Claude Code, the same default as Opus 5.5. The levels run low, medium, high, xhigh and max.
5. Know what fast mode bills. /fast works on Fable 5.2, but the fast mode guide is plain about who pays: “For Claude Code users on subscription plans (Pro/Max/Ultra/Team/Enterprise), fast mode is available via usage credits only and not included in the subscription rate limits.” That holds on Max and Ultra, and during Pro’s included week.
6. Watch scripts and scheduled runs on Pro. In an interactive session, Claude Code can ask before Fable draws on usage credits. With the -p flag, and in apps built on the Agent SDK that don’t show the prompt, it “never asks for consent. When a Fable request there would bill to usage credits, Claude Code bills it without asking.”
7. If you used best. From 2.1.321, the best alias “resolves to the same model as opus.” Earlier versions pointed it at Fable where Fable was available.

In the Claude apps on the web, desktop and phone, select “Fable 5.2” from the model picker. Cowork needs the latest Claude Desktop.

How good it is

Anthropic says the two models “are a step above Opus 5.5 in almost all the capabilities we evaluated.” These rows are from its own table:

Test	Fable 5.2	Fable 5.1	Opus 5.5	GPT-6 Astra	Gemini 4 Carbon
FrontierCode 2.0, agentic coding	49.1%	32.4%	45.5%	42.8%	41.1%
FrontierSWE v2, agentic coding	69.9%	56.3%	62.3%	65.5%	61.2%
AutomationBench, business workflows	51.4%	31.4%	42.5%	41.4%	53.9%
OSWorld 2.1, computer use	89.3%	83.5%	87.2%	79.5%	81.5%
Remote Labor Index, knowledge work	41.3%	17.9%	21.3%	20.8%	Not given
Humanity’s Last Exam, with tools	74.2%	65.6%	67.7%	57.2%	Not given

Anthropic ran each model at its highest available effort unless it says otherwise. On FrontierCode, both Fables ran at xhigh and Carbon at medium, and the Remote Labor Index score for Fable 5.2 is at high. Fable 5.2 doesn’t lead every row: Gemini 4 Carbon is ahead on AutomationBench, and on the AA-Briefcase knowledge-work test Fable 5.2 has the best rubric score, 66.1%, while Opus 5.5 has the higher Elo, 1800 to 1785.

One independent reading is already in. When we read Artificial Analysis’s leaderboard on November 9, Fable 5.2 at max effort scored 65 on its Intelligence Index, ahead of Gemini 4 Carbon (high) at 60 and Opus 5.5 (max) at 59. The cost per task shows the trade:

Model and effort	Intelligence Index	Cost per task
Claude Fable 5.2 (max)	65	$26.54
Claude Fable 5.2 (high)	62	$9.08
Gemini 4 Carbon (high)	60	$7.76
Claude Opus 5.5 (max)	59	$10.04
Claude Fable 5.2 (medium)	56	$3.80
Claude Opus 5.5 (medium)	48	$2.45

At high effort, Fable 5.2 beat Opus 5.5 at max for less per task. At max, a task cost more than two and a half times what it did on Opus 5.5 at max. At medium, where Claude Code starts both models, Fable 5.2 scored eight points higher than Opus 5.5 for about half again the cost.

What it blocks

Fable 5.2 keeps the safeguards that separate it from Mythos. Its classifiers “block high-risk cyber activities, including penetration testing, exploit generation, and binary-based vulnerability, as well as most tasks in research biology,” and “When a query is blocked, users will instead receive a response from Opus 4.8 for cybersecurity and Opus 5 for biology.” A “small set of topics” in AI development falls back to Opus 5 as well.

Anthropic says there are fewer of these switches than with Fable 5.1. In early testing across claude.ai, Claude Code and Cowork, “we observed about 55% fewer blocks with Fable 5.2 compared with Fable 5.1—about 60% fewer in cybersecurity and 40% fewer in biology.” Anthropic’s help article says the checks “review everything the model reads, not just your latest message,” so a file or a search result can trigger one.

Two more conditions. Fable models require 30-day data retention for safety monitoring by default, and aren’t available under zero data retention “unless expressly authorized by Anthropic.” And if your work is security or life sciences research, the route to the unblocked model is an application: every tier of the Cyber Verification Program now includes Mythos 5.2, and life scientists have their own program.

Where else you can use it

• GitHub Copilot. Generally available since November 9 for “Copilot Pro+, Max, Business, and Enterprise users,” and “billed at provider list pricing under usage-based billing.” On Business and Enterprise an administrator has to turn on its policy, which “is off by default.” GitHub says “Rollout will be gradual.”
• Cursor. Its price table lists Claude Fable 5.2 at $8 in and $40 out. Third-party models there are “available in Cursor at the model’s API list price” on Premium and above.
• The clouds. Anthropic’s release notes list the Claude API, Claude Platform on AWS, Amazon Bedrock, Google Cloud and Microsoft Foundry.
• Not Antigravity. Google’s models page lists Claude’s Opus and Sonnet models, and no Fable.

If you call the API, read the release notes before changing the model name. Fable 5.2 returns a 400 error in three cases that the notes and the model page call out: your code edits an earlier message and then sends back one of the model’s thinking blocks from a later turn; a request declares the older computer_20251124 tool, except on Amazon Bedrock; or a request forces tool use.

Fable 5.1, and the Fable 5.5 that didn’t come

Fable 5.1 isn’t going away yet. Anthropic’s deprecations page lists claude-fable-5-1 as Active, with retirement “Not sooner than September 1, 2027,” and claude-fable-5-2 as “Not sooner than November 9, 2027.” On a plan, both draw on the same Fable allowance. On the API, 5.1 now costs more than its replacement.

Since early October, leakers had called the next Fable “Fable 5.5,” and in late October “Claude Mythos 5.5” sat in hidden text on Anthropic’s own Mythos page. Anthropic never used either name in an announcement, and what arrived is 5.2. Our memo on the rumors has what was claimed and how each date turned out.

Using it beside your other agents

Enjoy runs Claude Code on your Claude plan, and its model list comes from Claude Code itself, so Fable 5.2 appears there once your Claude Code offers it, without an Enjoy update. If it’s missing, update Claude Code from the agent settings: it needs 2.1.321 or later. Enjoy remembers the last model you used with each agent, so after the update, open the settings and check which Fable is selected.

A conversation can use a different model for each message, so Opus 5.5 can do the routine steps and Fable 5.2 the hard one, in the same thread. Fable uses a plan’s limits faster, so watch them: the desktop app’s agent settings show how much of each limit is left, and when Claude runs out you can switch the conversation to another agent, such as Codex, which gets the messages it missed. Enjoy doesn’t add usage to a Claude plan, and it can’t give you a model your account doesn’t have, Mythos included. Whichever app you use, check whether usage credits are on in Claude’s settings before you pick Fable on Pro.

What Anthropic hasn’t said

• How much Fable half a week buys. It publishes no message or token counts, only that Fable uses limits “faster than other Claude models.”
• Whether Pro’s week comes back. The offer article says that when it ends, “Fable models go back to running on usage credits.”
• Whether a Fable 5.5, a Mythos 5.5 or a new Opus is coming. None is announced.
• When Fable 5.1 retires, beyond “Not sooner than September 1, 2027.”
• When Enterprise Frontier Safeguards opens up. The tool that lets Fable run with zero data retention is “now in private beta.”

We’ll update this memo when Anthropic fills these in.