Claude Fable 5.2 and Mythos 5.2 Bring Faster Technical Work

Anthropic released Claude Fable 5.2 for everyone and Claude Mythos 5.2 for vetted organizations, with up to 30% faster output and lower token prices. Here is

Anthropic’s Claude Fable 5.2 is now available to every user, and Claude Mythos 5.2 goes to organizations cleared through its Cyber Verification and Life Sciences programs. The two share one underlying model, generate output up to 30% faster than the 5.1 releases, and cost $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. For a business owner, the practical question is which model fits the technical work already on your plate, and what the token price means for the bill.

Business Owner’s Takeaway

  • Fable 5.2 is open to all users. It carries cybersecurity and biology safeguards that block some requests, which is the trade for general availability. Anthropic says those safeguards are as robust as Fable 5.1’s but produce fewer unnecessary blocks, and a third-party review from DataCamp puts the reduction at roughly 55% fewer safety blocks than Fable 5.1.
  • Mythos 5.2 is the same underlying model. Organizations pursuing beneficial applications in cybersecurity or biomedical research can apply for access to Mythos through Anthropic’s recently expanded Life Sciences and Cyber Verification Programs. Anthropic’s model documentation confirms Mythos 5.2 shares Fable 5.2’s specifications and pricing.
  • Long technical work is the target. Anthropic recommends starting with Claude Opus 5.5 for most coding and knowledge tasks and moving up to Fable or Mythos when a job is open-ended and hard.

Availability and Eligibility

Anthropic lists Claude Fable 5.2 as available to Pro, Max, Ultra, Team, and Enterprise users, and on the Claude Platform natively, in Amazon Web Services, Google Cloud, and Microsoft Foundry. The model documentation gives the context window as 1M tokens with a 128K max output, adaptive thinking that is always on, and a June 2026 knowledge cutoff. The docs also list the model as released November 9, 2026, with retirement not sooner than November 9, 2027.

Two constraints matter for anyone sending company data through the API. Anthropic states that Fable 5.2 requires 30-day data retention and is not available under zero data retention unless Anthropic expressly authorizes it. Adaptive thinking cannot be switched off, and depth is set with the effort parameter. Fast mode is available on the Claude Platform, Google Cloud, and Microsoft Foundry, and on Claude Code.

Verified Capabilities

The same release note and the System Card describe what the models did in prerelease testing: engineers’ software migrations, three electronic devices designed from scratch, two open problems in black-hole physics, progress on robotics evaluations, and a formalization of the proof of Fermat’s Last Theorem in 30 hours. The three devices were a home voice-assistant hub, a USB speakerphone, and a handheld virtual pet, each built with circuit boards that reached manufacturing readiness in under a day and each working without electrical modification.

On code, Anthropic says over 90% of changes merged into its core codebase are now written by Claude with human guidance, and that the model has been used to fix stubborn bugs in minutes and optimize performance across its services. A team of agents running an earlier version of Fable 5.2 migrated a critical production service from Python to Rust, and Anthropic reports that service is more efficient, more reliable, and starts about 26 times faster. In a separate experiment, one Anthropic engineer gave Fable 5.2 an empty folder and a request to design a prototype operating system. The model coordinated more than 1,000 subagents over eight days and produced a system that boots on a laptop, with its own kernel, disk, keyboard and network drivers, permanent file storage, windows, a login screen, an installer, and built-in Claude functionality.

Benchmark Limits

Anthropic publishes a comparison table where each model appears at its highest available effort and with safeguards on unless noted. Fable 5.2 scores 49.1% on FrontierCode 2.0 at Xhigh effort, 69.9% on FrontierSWE v2, 51.4% on AutomationBench, 89.3% on the offline OSWorld 2.1 set, 66.1% on the AA-Briefcase v1.1 rubric, and 41.3% on the Remote Labor Index at High effort. Mythos 5.2 records 74.6% on Terminal-Bench-Science 0.1, 76.3% on CyScenarioBench, and 89.6% on CVE Rediscovery.

The accuracy-versus-cost charts carry the conditions. On FrontierCode 2.0, Fable 5.2 at High effort scored 45.3% against Opus 5.5’s best of 45.5%, at roughly half the cost per task. At Medium effort on the WANDR agentic-search evaluation, it matched Opus 5.5’s top score at about 40% of the cost per task. On BenchCAD 2.1, which scores hardware designs against reference geometry and circuits, Fable 5.2 at Max effort reached 83.5 against 65.3 for Opus 5.5 and 44.4 for Fable 5.1. On Chartography, which tests chart and figure reading, Fable 5.2 at Medium effort scored 66.4% against 64.4% for Opus 5.5 at Max effort, at less than one-tenth of the cost per task. On Humanity’s Last Exam, Fable 5.2 at Low effort matched Opus 5.5’s top score at less than one-tenth of the cost per task.

Two limits in the table are worth noting. Scores on CyScenarioBench are based on unique tasks per lab, so the rows are not directly comparable. RoboDojo Sim shows the highest available effort and was run internally, and Interesting Engineering’s write-up describes it as a benchmark for controlling a simulated two-arm robot from camera images, with 64.9% for Fable 5.2 against 39.4% for Opus 5.5 and 29.8% for GPT-6 Astra. DataCamp notes that Fable 5.2 more than doubles Fable 5.1’s Remote Labor Index score, which points to where a point release still moves the needle.

Business Use Cases

Anthropic describes two roles for Fable 5.2: Advisor, where Haiku, Sonnet, or Opus models consult it to help solve harder problems, and Orchestrator, where it delegates simpler tasks to smaller models. For a business owner, the practical use is a long technical task that outlasts an ordinary chat window. Kie.ai’s side-by-side test of the same 3D cathedral prompt had Fable 5.2 finish in 1 hour 56 minutes against 3 hours 11 minutes for Opus 5.5, at $81.10 versus $45.97 for the run.

Early testers quoted by Anthropic include Vercel, Shopify, Datadog, Unity, Uber, Oracle, iGent, Snowflake, and Cognition. Joe Haddad, Member of Technical Staff at Vercel, said: “The model is in my head. It understands the intent behind every prompt I’ve given it.”

Pricing, Dated

As of Anthropic’s November 2026 announcement, standard pricing is $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. DataCamp calculates that as a 20% cut, applied to cache reads at $0.20 per million tokens and to batch processing as well. Fast mode runs at $16 per million input tokens and $80 per million output tokens for up to 2.5 times faster token generation. The docs list cache writes at $10 per million tokens for five minutes and $16 for one hour, plus a 50% Batch API discount on input and output. US-only inference runs at 1.1 times pricing. Check current pricing before committing a budget.

What to Do Now

  1. Pick one representative technical task your business actually runs and define what a good result looks like.
  2. Require a named human reviewer on any code, research, or design output before it reaches production.
  3. Run that same task on your current model and compare the two outputs side by side.
  4. Work out the total token cost of the task, including whether a fast-mode price fits the deadline.
  5. Check your data retention position against Anthropic’s 30-day requirement for Fable 5.2 before sending anything proprietary.
  6. Apply through the Cyber Verification or Life Sciences program if your organization does eligible work and you need Mythos 5.2.

FAQ

What Are Claude Fable 5.2 and Claude Mythos 5.2?

They are Anthropic’s models for demanding technical work, with improvements in coding, reasoning, and vision. They are the same underlying model: Fable 5.2 has cybersecurity and biology safeguards for general availability, while Mythos 5.2 is offered through Anthropic’s verification programs with fewer of those blocks.

How Much Do Claude Fable 5.2 and Mythos 5.2 Cost?

As of the November 2026 announcement, standard pricing is $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. Fast mode is $16 per million input tokens and $80 per million output tokens.

Who Can Access Claude Mythos 5.2?

Organizations pursuing cybersecurity or biomedical research apply through Anthropic’s Cyber Verification and Life Sciences programs. Mythos 5.2 shares Fable 5.2’s specifications and pricing, so there is no separate model price.

Related Coverage

This article reports on publicly available information from the sources listed below. We have no reason to believe it is inaccurate, but we have not independently verified every claim, and details may change as the story develops.

Sources


See our editorial disclaimer for how our articles are produced.

🤖
Is your business visible to AI assistants?

Run a free scan to see your AI Visibility Score, SEO rating, and local citation accuracy.

Check Your Score →