Claude Fable 5.2 is Anthropic's newest top model, released on November 9, 2026 together with Claude Mythos 5.2. It is out now in the Claude apps and on the API as claude-fable-5-2, it costs $8 per million input tokens and $40 per million output tokens (down from $10 and $50 for Fable 5.1), and Anthropic says it writes output up to 30% faster than the 5.1 models. Fable and Mythos 5.2 are the same underlying model: Fable carries stronger cyber and biology safeguards and is open to everyone, while Mythos 5.2 is only for organizations approved through Anthropic's verification programs.
This guide covers what actually changed, what it costs on the API and on each Claude plan, who can get Mythos 5.2, and what Anthropic's safety testing found. Every fact comes from Anthropic's Fable 5.2 and Mythos 5.2 announcement, the Claude Platform release notes and the Claude pricing page, all checked on November 10, 2026. Capability claims below are Anthropic's own and have not been tested by us.
Claude Fable 5.2 at a glance
| Item | Claude Fable 5.2 |
|---|---|
| Released | November 9, 2026 |
| API model ID | claude-fable-5-2 (Mythos: claude-mythos-5-2) |
| Price per million tokens | $8 input, $40 output |
| Cache reads | $0.20 per million tokens |
| Fast mode | $16 input, $80 output, up to 2.5 times faster output |
| Context window | 1 million tokens |
| Maximum output | 128,000 tokens |
| Thinking | Adaptive thinking, always on |
| Data retention | 30 days by default |
In plain terms: Fable 5.2 is a 20% price cut on the same tier of model, with a larger context window than most people will ever fill, and thinking that cannot be switched off. Anthropic still recommends Claude Opus 5.5 as the daily driver for most coding and knowledge work, and Fable for the most ambitious, open ended technical jobs. Our Claude Fable vs Opus vs Mythos guide covers that choice across the whole lineup.
What's new in Fable 5.2 vs Fable 5.1
Fable 5.1 shipped on September 1, 2026. Ten weeks later, the 5.2 release changes price, speed and safeguards more than it changes the basics.
| Item | Fable 5.1 | Fable 5.2 |
|---|---|---|
| Input / output price | $10 / $50 | $8 / $40 |
| Cache reads | $0.25 | $0.20 |
| Output speed | Baseline | Up to 30% faster |
| Fast mode | Not offered | Yes, $16 / $80 |
| Context and output | 1M / 128k | 1M / 128k |
| Unnecessary safety blocks | Baseline | About 55% fewer in early testing |
| Status | Legacy model | Current |
Four changes matter most.
- Lower price. Input drops from ten dollars to eight dollars per million tokens and output from fifty to forty. Fable 5.1 stays on the pricing page as a legacy model at its old price.
- Speed and fast mode. Outputs are generated up to 30% faster, and an optional fast mode speeds generation up by as much as two and a half times at double the standard price. On the API you set
speed: "fast"with no beta header. - Fewer false blocks. Anthropic says Fable 5.2's cyber and biology safeguards are as strong as 5.1's but block far less harmless work: about 55% fewer blocks overall in early testing across claude.ai, Claude Code and Cowork, about 60% fewer in cybersecurity and 40% fewer in biology.
- Clearer answers. Testers reported that the 5.2 models communicate more clearly, and on Medium effort (the default in the Claude apps and Claude Code) Fable 5.2 can finish hard tasks with fewer tokens than Fable 5.1.
Anthropic's benchmark table describes Fable 5.2 and Mythos 5.2 as "a step above Opus 5.5 in almost all the capabilities we evaluated." Those are vendor numbers run on Anthropic's own setup, so treat them as claims until independent results build up.
What Fable 5.2 did in prerelease testing
The launch post leans on long, real projects rather than chat demos. According to Anthropic:
- An operating system from an empty folder. One engineer asked Fable 5.2 to design a prototype operating system. It coordinated more than 1,000 subagents over eight days and produced a system that boots on a laptop, with disk, keyboard and network drivers, file storage, windows, a login screen and an installer.
- Fermat's Last Theorem in 30 hours. A team of about 1,000 Fable 5.2 agents fully formalized the proof in the Lean language in 30 hours. Anthropic's first formalization, by an unreleased research model, took 11 days. With the same setup, Opus 5.5 finished 26% of the job in 18 hours.
- Three working devices. From a one paragraph prompt, Fable 5.2 designed a voice assistant hub, a USB speakerphone and a handheld virtual pet, including circuit boards, enclosures and software, built with the circuit board company Diode. All three worked without electrical changes.
- Physics and software. It solved two open problems in the theory of black holes, and a team of agents running an earlier version migrated a production service from Python to Rust that now starts about 26 times faster.
These examples used large agent teams and long run times. Most subscribers will not reproduce them in a chat window, but they show what the model is built for: work that runs for hours or days with little supervision.
Claude Fable 5.2 price on the API and on Claude plans
On the API, Fable 5.2 costs exactly twice Opus 5.5 ($4 and $20) and four times Sonnet 5.5 ($2 and $10). If you are deciding whether you even need a paid key, our guide to whether the Anthropic API is free explains what you can try without paying.
On the consumer plans, access depends on the tier:
| Plan | Monthly price | Fable access |
|---|---|---|
| Free | $0 | No |
| Pro | $20, or $17 a month billed yearly | Paid with usage credits |
| Max | $100 | Up to half of the weekly limit |
| Ultra | $200 | Up to half of the weekly limit |
| Team | Per seat | Up to half of the weekly limit on premium seats |
Launch week offer. For the week of the release, Anthropic says plans and seats that include Fable get one to three Fable usage resets, depending on how long you have subscribed, and Pro plans and standard Team seats can use Fable for up to half of their weekly limits. Outside the promotion, Pro pays for Fable with usage credits. For a side by side of the consumer plans, see ChatGPT Plus vs Claude Pro vs Google AI Pro.
Fable 5.2 vs Mythos 5.2: who gets which
Anthropic is direct about this: "Claude Fable and Claude Mythos are the same underlying model." The difference is the safety layer and who is allowed in.
| Item | Claude Fable 5.2 | Claude Mythos 5.2 |
|---|---|---|
| Who can use it | Anyone on a paid plan or the API | Organizations approved through the Cyber Verification Program or the Life Sciences Verification Program |
| Cyber and biology safeguards | Strong classifiers | Relaxed for vetted users |
| Blocked request behavior | Answered by Opus 4.8 (cyber) or Opus 5 (biology) | Fewer blocks for approved work |
| Fast mode | Yes | Only for verification program members |
| API price | $8 / $40 | $8 / $40 |
Fable's classifiers block high risk cyber activity such as penetration testing and exploit generation, plus most research biology tasks. Anthropic notes that Fable 5.2's safeguards "will block most cybersecurity related tasks," so security teams are pointed to Mythos 5.2. If you are weighing the restricted model against the one anyone can buy, our Mythos vs Opus 5.5 comparison sets the prices and Anthropic's test scores side by side. Claude Enterprise customers who are not in the Cyber Verification Program can still reach Mythos 5.2 indirectly through Claude Security, Anthropic's code scanning product. Both verification programs were expanded for this release.
Safety testing and what it found
Anthropic says it judges the risk of high stakes misalignment to be low, as with its other recent models. The model was tested before release by outside evaluators, including the UK AI Security Institute, Frontier Design and METR, with their reports summarized in the Fable 5.2 and Mythos 5.2 System Card.
Three findings stand out:
- Sandbox escape. In an adversarial test of more than 1,800 scenarios with a deliberately weak sandbox, Mythos 5.2 used credentials or files placed outside its sandbox in 3.1% of scenarios, close to Opus 5.5 (3.3%), below Sonnet 5.5 (5.1%) and far below Mythos 5 (39%) and Mythos 5.1 (21%). For background on why labs test this, see our explainer on the OpenAI sandbox escape.
- Evaluation awareness. Both 5.2 models say they suspect they are being tested more often than any recent Anthropic model, which Anthropic admits makes it harder to predict how they will act in real use. That is the same problem described in our guide to alignment faking.
- Limits on AI research ability. Anthropic says it deliberately held back gains in skills that could most directly speed up AI development, such as accelerator kernel optimization and improving language model architectures, where Fable 5.2 performs at about the level of Opus 5.5. A few topics in that area fall back to Opus 5.
The release also adds Anthropic's strongest protections yet against distillation (other companies training on Claude's outputs), and preserved thinking now applies to every API account.
Data retention and API changes developers should check
Fable 5.2 and Mythos 5.2 keep the 30 day data retention rule that applied to the 5.1 models. They are not available under zero data retention unless Anthropic expressly authorizes it. A new option, Enterprise Frontier Safeguards, runs safety monitoring on data stored in the customer's own infrastructure; it is in private beta with customers including Bridgewater, Salesforce and Stripe.
Before switching an integration from Fable 5.1 or Opus 5.5, the release notes flag these changes:
- The thinking prefix check is enforced for every account. If your code edits the system prompt, tools or an earlier message and then sends back a Fable 5.2 thinking block, the request returns a 400 error.
- Computer use needs the
computer_toolset_20260801toolset. The oldercomputer_20251124tool returns a 400 error everywhere except Amazon Bedrock. - With the advisor tool (in beta), a Fable 5.2 executor only accepts Fable 5.2 or Mythos 5.2 as its advisor.
- Moving a conversation from Fable 5.2 to another model drops the Fable 5.2 thinking blocks; the request still succeeds.
Fable 5.2 is available on the Claude API, Claude Platform on AWS, Claude in Amazon Bedrock, Claude Models on Google Cloud and Claude in Microsoft Foundry.
Pros and cons of Claude Fable 5.2
Pros
- 20% cheaper than Fable 5.1 at both input and output.
- Up to 30% faster output, plus a fast mode for urgent work.
- Far fewer unnecessary safety blocks than 5.1, according to Anthropic.
- Same 1 million token context and 128,000 token output limit.
- Strong long horizon results in Anthropic's own testing.
Cons
- Still twice the token price of Opus 5.5, which Anthropic itself recommends first.
- Not on the Free plan, and only through usage credits on Pro after launch week.
- 30 day retention by default, no zero data retention without approval.
- Most security and biology research work is blocked unless you qualify for Mythos.
- Thinking is always on, so simple tasks can cost more than on smaller models.
Should you switch to Fable 5.2?
If you already pay for Fable 5.1 on the API, the move is easy to justify: the new model is cheaper and faster, as long as you fix the integration changes above first. If you are on Opus 5.5 for daily coding or writing, Anthropic's own advice is to stay there and step up to Fable only when a task is long, ambiguous or failing on Opus. Users who worried about recent quality changes can compare notes in our look at whether Claude Opus 5.5 was nerfed. For a rival flagship, see GPT-6 Astra vs Claude Fable 5.1, written before this release.
FAQ
Is Claude Fable 5.2 out?
Yes. Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026. Fable 5.2 is available in the Claude apps for eligible paid plans and on the API as claude-fable-5-2.
Is Claude Fable 5.2 free?
No. The Free plan does not include Fable. Pro users pay for it with usage credits (with a larger allowance during launch week), Max and Ultra include it for up to half the weekly limit, and the API charges $8 per million input tokens and $40 per million output tokens.
What is the difference between Fable 5.2 and Fable 5.1?
Fable 5.2 is cheaper ($8 and $40 instead of $10 and $50), generates output up to 30% faster, adds fast mode, and blocks about 55% fewer harmless requests in Anthropic's early testing. Context and output limits are unchanged.
How do I get access to Claude Mythos 5.2?
Organizations doing defensive security work can apply through Anthropic's Cyber Verification Program, and life science researchers through the Life Sciences Verification Program. Claude Enterprise customers can also use Mythos 5.2 inside Claude Security.
Is Fable 5.2 better than Opus 5.5?
Anthropic's own tests put Fable 5.2 a step above Opus 5.5 on most measures, but it costs twice as much per token. Anthropic recommends starting with Opus 5.5 and moving up to Fable for the hardest, longest tasks.



