Claude Fable 5.2 and Mythos 5.2: Release, Price and Benchmarks

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, calling them its "most capable models for daunting technical work," with gains in coding, finance, and the sciences. Fable 5.2 is available immediately to all Claude users and on every major cloud platform at $8 per million input tokens and $40 per million output tokens; Mythos 5.2 — described as the same underlying model with different safeguards — remains gated behind Anthropic's verification programs for cybersecurity and life-sciences work.

What Anthropic announced

In a post on its newsroom, Anthropic introduced Claude Fable 5.2 and Claude Mythos 5.2 as updates built for what it calls "daunting technical work" — long-horizon coding projects, agentic business workflows, and scientific research. The company frames the release as an upgrade over the Fable 5.1 generation it shipped in October rather than a successor to Opus 5.5, its general-purpose flagship. If you're catching up on the earlier model, our explainer on what Claude Fable 5.1 actually is covers how that version positioned itself before this week's update.

Anthropic's naming keeps the Fable/Mythos split it introduced last month: Fable is the publicly available model, and Mythos is a restricted sibling intended for sensitive, high-stakes domains. According to Anthropic, "Fable and Mythos are the same underlying model," but Fable ships with stronger cyber and biological safeguards baked in by default. Mythos 5.2, by contrast, is tuned and vetted for use inside Anthropic's verification programs, where organizations with a legitimate need for less-constrained cyber or biology capability can apply for access.

Pricing: what Fable 5.2 and Mythos 5.2 cost

Anthropic's pricing page and the launch announcement both list the same standard rate for Fable 5.2: $8 per million input tokens and $40 per million output tokens. That is double Opus 5.5's listed rate of $4 input / $20 output per million tokens, positioning Fable as Anthropic's most expensive model to run at scale. The announcement states that Mythos 5.2 is billed at the same standard rate as Fable 5.2, though Anthropic's pricing page does not break Mythos out as its own line item — unsurprising, since access to Mythos is gated rather than self-serve.

Both Fable 5.2 and Opus 5.5 also offer a fast mode, which Anthropic says generates tokens up to 2.5x faster at roughly double the standard price — $16 per million input tokens and $80 per million output tokens for Fable 5.2. Fast mode is available through Claude Code, the Claude Platform, Claude Models on Google Cloud, and Microsoft Foundry; a fast-mode version of Mythos 5.2 exists too, but Anthropic limits it to organizations already inside its verification programs.

Anthropic's pricing page notes that plans bundle context windows of up to 1 million tokens depending on the model, but neither the pricing page nor the announcement publishes an exact context-window figure specific to Fable 5.2 or Mythos 5.2.

Benchmarks: how Fable 5.2 compares

Anthropic's announcement leans heavily on first-party benchmark results, comparing Fable 5.2 and Mythos 5.2 against its own Opus 5.5 as well as unnamed rivals it labels "GPT-6 Astra" and "Gemini 4 Carbon." Because these are Anthropic's own reported numbers rather than independently reproduced scores, they should be read as a vendor's self-reported comparison rather than a neutral benchmark.

Anthropic's biggest reported jumps over the prior Fable 5.1 generation show up in science and automation work: FrontierSWE v2 climbs from 56.3% to 69.9%, Terminal-Bench-Science 0.1 from 52.6% to 74.6%, and AutomationBench from 31.4% to 51.4%. The company also cites an internal Elo-style rating — 1785 for Fable 5.2 versus 1800 for Opus 5.5 — without naming the underlying evaluation, so it's not directly comparable to the named benchmarks above.

Notably, the announcement does not publish a SWE-bench Verified score or a dedicated finance benchmark for Fable 5.2, despite finance being named in the launch framing. The closest finance-adjacent figure Anthropic cites is AA-Briefcase v1.1, a rubric-graded knowledge-work benchmark, where Fable 5.2 scores 66.1% against Opus 5.5's 61.2%.

Fable 5.2 vs. Mythos 5.2 vs. Opus 5.5

Anthropic is explicit that Fable 5.2 and Mythos 5.2 are not meant to replace Opus 5.5 as most people's default model. The company describes Opus 5.5 as its "strongest daily driver" for most coding and knowledge-work tasks, reserving Fable and Mythos for "the most ambitious work" — long, difficult coding problems, large agentic workflows, and scientific or security research where the extra capability (and roughly 2x the price) is worth it. For a side-by-side on how the previous Fable generation stacked up against Opus 5.5, see our earlier comparison of Opus 5.5 vs. Fable 5.1, which lays out the same trade-off Anthropic is reiterating with 5.2.

The practical difference between Fable and Mythos isn't capability — Anthropic says they're the same underlying model — but access and safeguards. Fable 5.2 defaults to stronger built-in cyber and biosecurity guardrails so it can be deployed broadly. Mythos 5.2 trades some of those default guardrails for raw capability in narrow, vetted contexts: cybersecurity research and life-sciences research, specifically. That's consistent with how Anthropic has described Mythos generally; our earlier piece on what Claude Mythos is and why it's restricted goes into the access model in more depth.

Availability: where you can use them

Fable 5.2 is live now across Anthropic's own surfaces and third-party cloud platforms. Per the announcement, it's available on the Claude apps to all users, and through the API on Anthropic's own platform as well as Amazon Web Services, Google Cloud, and Microsoft Azure. Paid Claude subscribers get access to Fable within their plan, with Anthropic noting that paid plans receive one to three usage resets depending on subscription length, and that Fable can account for up to 50% of weekly usage limits on Pro and standard Team seats.

Mythos 5.2 is handled differently. Access runs through two gated programs: the Cyber Verification Program for security research organizations, and a parallel Life Sciences Verification Program for biology and life-sciences work. Outside of those programs, Claude Enterprise customers can use Mythos 5.2 indirectly through Claude Security, Anthropic's security product, which the company says runs on Mythos 5.2 under the hood.

Model and deployment details, including the Fable 5.2 API identifier, are documented on Anthropic's developer platform under its models overview page, which developers migrating existing integrations should check against their current model IDs.

What's actually new in coding, finance, and the sciences

Anthropic's framing for this release leans on three use cases: coding, finance, and the sciences. On coding, the gains are backed by two named benchmarks (FrontierCode 2.0 and FrontierSWE v2), both of which show double-digit-point improvements over Fable 5.1 and a lead over Opus 5.5. Anthropic also points to gains on OSWorld 2.1, a computer-use benchmark, and on agentic search (WANDR), suggesting the improvements extend to models that operate tools and interfaces autonomously rather than just writing code in isolation.

On finance, the announcement is thinner on specifics. Anthropic doesn't cite a named finance benchmark; the strongest related evidence is Fable 5.2's score on AA-Briefcase v1.1, a general knowledge-work rubric, and on the Remote Labor Index, where it scores 41.3% against Opus 5.5's 21.3%. Readers looking for a finance-specific, independently verified benchmark won't find one in this announcement.

The sciences claim is better substantiated. Terminal-Bench-Science 0.1 scores jump substantially for both Fable 5.2 (74.6%) and Mythos 5.2 (75.3%) over Fable 5.1's 52.6%, and Mythos 5.2's biology-adjacent protein-design score (85.1% on a 24-hour de novo binder design task) edges out Opus 5.5's 82.6% figure — though Anthropic notes the Opus 5.5 number was measured "without safeguards," muddying a direct comparison.

Safety: the system card and verification programs

Alongside the launch, Anthropic published a system card for Claude Fable 5.2 and Mythos 5.2, its standard practice for frontier releases. The existence of a dedicated system card, plus the two-tier Fable/Mythos access structure and the Cyber and Life Sciences Verification Programs gating Mythos, signals that Anthropic is treating this release's cyber and biology capability jumps — reflected in benchmarks like CyScenarioBench and CVE Rediscovery — as meaningful enough to warrant restricted deployment rather than a blanket public launch for the more capable configuration.

Frequently asked questions

When did Claude Fable 5.2 and Mythos 5.2 launch?
Anthropic announced both models on November 9, 2026.

How much does Claude Fable 5.2 cost?
$8 per million input tokens and $40 per million output tokens at the standard rate, or $16 / $80 per million tokens in fast mode, per Anthropic's pricing page.

What's the difference between Fable 5.2 and Mythos 5.2?
Anthropic says they're the same underlying model. Fable 5.2 ships with stronger default cyber and biosecurity safeguards for broad availability; Mythos 5.2 is access-gated through verification programs for vetted cybersecurity and life-sciences work.

Is Claude Fable 5.2 better than Claude Opus 5.5?
Anthropic's own benchmarks show Fable 5.2 ahead of Opus 5.5 on most reported tests, but the company still recommends Opus 5.5 as the default "daily driver" for most coding and knowledge-work tasks, positioning Fable for the hardest jobs.

Who can access Claude Mythos 5.2?
Organizations vetted through Anthropic's Cyber Verification Program or Life Sciences Verification Program, plus Claude Enterprise customers using Claude Security, which runs on Mythos 5.2.

Is Claude Fable 5.2 available on AWS, Google Cloud, and Azure?
Yes — Anthropic says Fable 5.2 is available today across its own platform and on Amazon Web Services, Google Cloud, and Microsoft Azure.

What's next

Anthropic hasn't given a timeline for when — or whether — Mythos 5.2 might open beyond its current verification programs, and the company has not published independently verified benchmark numbers to sit alongside its own figures. Developers evaluating a migration should watch Anthropic's models overview documentation for the finalized Fable 5.2 API identifier and any context-window specifications, and should expect third-party benchmark sites and reviewers to begin publishing their own comparisons against Opus 5.5 and competing frontier models in the coming weeks. Pandromeda will update this story if Anthropic publishes additional pricing, context-window, or access details.

Written byTheo Park

Theo Park runs the AI desk at Pandromeda. He follows model launches from the frontier labs and the open-weight community, tracks the assistants and developer tools built on them, and explains what each release changes on pricing, capability and safety. His reporting leans on primary sources: model cards, technical reports, API documentation and the companies' own announcements.

More from Theo ParkAll AI coverage