Anthropic Releases Claude Fable 5.2 and Mythos 5.2 Models

We’ve upgraded Claude Fable and Claude Mythos—models that are designed to tackle the most daunting technical challenges.

Claude Fable 5.2 and Claude Mythos 5.2 show improvements in areas like coding, reasoning, and vision. During prerelease testing they’ve helped our engineers complete otherwise-infeasible software migrations, designed three working electronic devices from scratch, solved two open problems in the physics of black holes, made substantial progress on robotics evaluations, and formalized the proof of Fermat’s Last Theorem in 30 hours—among many other applications we’ll discuss below.

Claude Fable and Claude Mythos 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. Organizations pursuing beneficial applications in cybersecurity or biomedical research can apply for access to Mythos through our recently expanded Life Sciences and Cyber Verification Programs.

Testers have found that Fable 5.2 and Mythos 5.2 communicate more clearly than past Fable and Mythos models. They generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1, and they’re available at a lower price: $8 per million input tokens and $40 per million output tokens.

Fable 5.2 and Mythos 5.2 also come with an optional fast mode, which speeds up token generation by up to 2.5 times. Fast mode is available for $16 per million input tokens and $80 per million output tokens.

Safety and alignment. To help ensure safety and alignment keep step with growing model capabilities, we ran extensive prerelease safety testing on the model that underlies Fable 5.2 and Mythos 5.2. This included developing novel evaluations that give deeper insight into specific concerning behaviors like sandbox escapes. We believe high-stakes risks from misalignment remain low, and the model performed as well as, or better than, our recent models on most alignment measures. Fable 5.2’s cyber and biology safeguards are as robust as those of Fable 5.1’s but produce fewer unnecessary blocks, and we have implemented our strongest-ever protections against distillation attacks.

We describe more about our safety testing below, and full details can also be found in the Claude Fable 5.2 and Mythos 5.2 system card.

When to use Claude Fable 5.2 and Mythos 5.2. Fable and Mythos are best for the most ambitious, open-ended technical work. For most coding and knowledge tasks, we recommend that users start with Claude Opus 5.5, our strongest daily driver, and then switch to a different model as required, depending on the scope and complexity of their work—upward to Fable or Mythos, or downward to Claude Sonnet 5.5 or Claude Haiku 5.5. Fable 5.2 can also play the role of Advisor (where Haiku, Sonnet, or Opus models consult it to help solve harder problems) or Orchestrator (where it delegates simpler tasks to smaller models).

Evaluating Claude Fable 5.2 and Mythos 5.2

Benchmark performance. The benchmarks table below shows that Claude Fable 5.2 and Mythos 5.2 are a step above Opus 5.5 in almost all the capabilities we evaluated:

| | Fable 5.2 | Fable 5.1 | Opus 5.5 | GPT-6 Astra | Gemini 4 Carbon | | --- | --- | --- | --- | --- | --- | | Agentic codingFrontierCode 2.0 | | | | | | | Agentic codingFrontierCode 2.0 | 49.1%Xhigh effort | 32.4%Xhigh effort | 45.5% | 42.8% | 41.1%Medium effort | | Agentic codingFrontierSWE v2 | | | | | | | Agentic codingFrontierSWE v2 | 69.9% | 56.3% | 62.3% | 65.5% | 61.2% | | Business workflowsAutomationBench | | | | | | | Business workflowsAutomationBench | 51.4% | 31.4% | 42.5% | 41.4% | 53.9% | | Agentic scientific researchTerminal-Bench-Science 0.1 | | | | | | | Agentic scientific researchTerminal-Bench-Science 0.1 | 74.6%75.3% (Mythos 5.2) | 52.6% | 58.7% | 68.1% | 64.8% | | CybersecurityCyScenarioBench† | | | | | | | CybersecurityCyScenarioBench† | 76.3%Mythos 5.2 | 61.7%Mythos 5.1 | 67.6%without safeguards | 61%GPT-6 Cyber | — | | CybersecurityCVE Rediscovery | | | | | | | CybersecurityCVE Rediscovery | 89.6%Mythos 5.2 | 78.3%Mythos 5.1 | 81.3%without safeguards | — | — | | Protein designDe novo binder design (24 hr) | | | | | | | Protein designDe novo binder design (24 hr) | 85.1%Mythos 5.2 | 79.1%Mythos 5.1 | 82.6%without safeguards | — | — | | Agentic searchWANDR (relaxed score) | | | | | | | Agentic searchWANDR (relaxed score) | 82.1% | 68.7% | 72.3% | — | — | | Computer useOSWorld 2.1 (offline set, partial) | | | | | | | Computer useOSWorld 2.1 (offline set, partial) | 89.3% | 83.5% | 87.2% | 79.5% | 81.5% | | RoboticsRoboDojo Sim‡ | | | | | | | RoboticsRoboDojo Sim‡ | 64.9% | 27.6% | 39.4% | 29.8% | — | | Knowledge workAA-Briefcase v1.1 | | | | | | | Knowledge workAA-Briefcase v1.1 | 66.1%rubric | 61.5%rubric | 61.2%rubric | 52.0%rubric | 61.8%rubric | | 1785Elo | 1669Elo | 1800Elo | 1569Elo | 1611Elo | | | Knowledge workRemote Labor Index (RLI) | | | | | | | Knowledge workRemote Labor Index (RLI) | 41.3%High effort | 17.9% | 21.3% | 20.8% | — | | Multidisciplinary reasoningHumanity’s Last Exam (with tools) | | | | | | | Multidisciplinary reasoningHumanity’s Last Exam (with tools) | 74.2% | 65.6% | 67.7% | 57.2% | — | | Visual chart recognitionChartography (no tools) | | | | | | | Visual chart recognitionChartography (no tools) | 73.4% | 44.8% | 64.4% | 71.0% | 72.5% |

For details on evaluation methodology, see the Claude Fable 5.2 and Claude Mythos 5.2 System Card. Scores for Fable 5.2, Fable 5.1, and Opus 5.5 show results with safeguards on unless otherwise noted. All scores show each model’s highest-available effort unless otherwise noted. † On CyScenarioBench, each lab’s scores are based on unique tasks. ‡ RoboDojo Sim shows the highest-available effort and was run internally.

Effort and efficiency. On Medium effort (the default in Claude apps and Claude Code), Fable 5.2 can accomplish challenging tasks with fewer tokens than Fable 5.1. In early testing on the 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. At Xhigh effort, it also scored the highest of any released model tested.

Agentic coding: FrontierCode 2.0Agentic search: WANDRHardware engineering: BenchCAD 2.1Multidisciplinary reasoning: Humanity’s Last ExamVisual chart recognition: ChartographyAgentic coding: FrontierCode 2.0Agentic search: WANDRHardware engineering: BenchCAD 2.1Multidisciplinary reasoning: Humanity’s Last ExamVisual chart recognition: ChartographyFrontierCode 2.0Accuracy vs. cost

• Fable 5.2
• Opus 5.5
• Fable 5.1
• GPT-6 Astra
• Gemini 4 Carbon

10203040500Score (%)125102050Cost per task (USD, log scale)LowMedHighXhighMax

FrontierCode measures whether a model’s code changes would be merged by the maintainers of open-source repositories. At High effort, Fable 5.2 scores about the same as Opus 5.5’s best score (45.3% vs. 45.5%) for roughly half the cost per task, and ahead of GPT-6 Astra’s best (42.8%) at about the same cost. At Xhigh effort, Fable 5.2 reaches 49.1%, the highest score of any released model we’ve tested.

WANDR (relaxed score)Accuracy vs. cost

• Fable 5.2
• Opus 5.5
• Fable 5.1

304050607080900Score (%)125102050Cost per attempt (USD, log scale)LowMedHighXhighMax

Perplexity’s WANDR benchmark measures agents on large data-collection tasks. At Medium effort, Fable 5.2 matches Opus 5.5’s top score for about 40% of the cost per task.

BenchCAD 2.1 (with tools)Accuracy vs. cost

• Fable 5.2
• Opus 5.5
• Fable 5.1

020406080Score (0–100)0.030.100.30131030Cost per attempt (USD, log scale)LowMedHighXhighMax

BenchCAD 2.1 measures hardware engineering, specifically how closely a model’s designs match reference geometry and circuits, and is scored 0–100 with partial credit. At Max effort, Fable 5.2 scores 83.5, ahead of Opus 5.5 (65.3) and Fable 5.1 (44.4).

Humanity’s Last Exam (with tools)Accuracy vs. cost

• Fable 5.2
• Opus 5.5
• Fable 5.1

506070800Score (%)0.100.200.50125Cost per attempt (USD, log scale)GPT-6 Astra (Max): 57.2%LowMedHighXhighMax

Humanity’s Last Exam (HLE) is a test of expert-level academic knowledge and reasoning. At Low effort, Fable 5.2 matches Opus 5.5’s top score for less than one-tenth of the cost per task.

Chartography (no tools)Accuracy vs. cost

• Fable 5.2
• Opus 5.5
• Fable 5.1

3040506070800Score (%)0.020.050.100.200.501Cost per attempt (USD, log scale)GPT-6 Astra (Max): 71.0%Gemini 4 Carbon: 72.5%LowMedHighXhighMax

Chartography tests how accurately a model can read charts and figures. At Medium effort, Fable 5.2 scores 66.4%, ahead of Opus 5.5 at Max effort (64.4%) for less than one-tenth of the cost per task.

Coding

Over 90% of the changes merged into Anthropic’s core codebase are now written by Claude (with human guidance). We’ve used Fable 5.2 to fix stubborn bugs in minutes, optimize performance across all of our services, and complete high-value software projects that would otherwise have been too slow or too costly to attempt. A team of agents running an earlier version of Fable 5.2 migrated a critical production service from Python to Rust—the service is now more efficient, more reliable, and starts about 26 times faster. We’ve found that Fable 5.2 exercises strong engineering judgement and can produce high-quality results from even ambiguous requests.

As an experiment, one Anthropic engineer gave Fable 5.2 the task of designing a prototype operating system. Starting from an empty folder, the model coordinated over 1,000 subagents, working around the clock for eight days writing, testing, and reviewing code. It produced a working operating system that boots on a laptop, with its own core that runs the machine, drivers for the disk, keyboard, and network, permanent file storage, windows, a login screen, an installer, and built-in Claude functionality (see the video below).

Early testers have been impressed at how successfully Fable 5.2’s coding skills can be applied to a wide range of tasks:

VercelShopifyDatadogUnityUberOracleiGentSnowflakeCognitionVercelShopifyDatadogUnityUberOracleiGentSnowflakeCognitionQuote

“The model is in my head. It understands the intent behind every prompt I’ve given it. I haven’t had to say ‘No, I meant…’ a single time. Projects we’ve been pushing along with other models for weeks got finished and approved by our teammates in a few days with Claude Fable 5.2, and the humans reviewing its code had less to say.”

CompanyVercelAuthorJoe Haddad, Member of Technical StaffQuote

“Claude Fable 5.2 is great for taking a loose idea and refining it. Its answers are really well-considered, and within a few turns I can almost always say ‘ok, go do that,’ and trust that it’ll complete it with little intervention. It feels like working with another staff-level engineer.”

CompanyShopifyAuthorTom Pritchard, Staff EngineerQuote

“You can just throw Claude Fable 5.2 at new, ambiguous systems it has never seen before, and it will figure out how to actually work within them. It’s almost like an incident commander that can send out faster models to get quick signals and coordinate, which removes that burden from me as an operator.”

CompanyDatadogAuthorMatt Lee, Senior Staff EngineerQuote

“Claude Fable 5.2 can complete tasks that previous models struggled with. The clearest examples so far are building higher-quality 3D scenes from scratch and rebuilding an experience from a visual reference. It’s very strong on the harder game-dev tasks, where it keeps iterating and improving instead of getting stuck.”

CompanyUnityAuthorSam Zhang, Creative TechnologistQuote

“I had Claude Fable 5.2 run a 10–12 hour security audit. It sent out about 170 reviewer agents and came back with a write-up, fix pull requests and a policy proposal. It works until the job is actually complete instead of stopping at a plausible checkpoint. Its self-correction and honesty about gaps impressed me. At High effort, I’d use this over what I run today for long unattended audits and doc-heavy work.”

CompanyUberAuthorAndrew Mager, Staff Engineer, Agentic SystemsQuote

“In one test combining country debt summaries, repayment history, and regional benchmarks, Claude Fable 5.2 generated SQL that returned the expected results. The SQL generated by two other models also ran successfully, but incorrectly summed country-level summary amounts, producing values 5–8x the expected amounts. Getting these business calculations right is why we see value in using Fable 5.2 for enterprise SQL generation.”

CompanyOracleAuthorLiyu Gong, Senior Manager of Machine LearningQuote

“In a single session, Claude Fable 5.2 built a hypervisor, the software that walls off virtual computers sharing one machine. Its core is mathematically proven to keep them apart. More than 3,500 machine-checked proofs cover every case, not just the ones you test. It’s the first model I’ve seen break a problem this large into pieces it can prove, fit them together, and keep taking on more ambitious work. A very clear upgrade from Fable 5.1.”

CompanyiGentAuthorSean Ward, Co-Founder and CEOQuote

“Claude Fable 5.2 is the strongest model we have measured on data-eng-bench, our open-source benchmark of 103 data-engineering tasks. It solved tasks consistently across repeated runs and passed every interactive coding task we gave it.”

CompanySnowflakeAuthorBala Kasiviswanathan, VP of AI and Developer ExperienceQuote

“Claude Fable 5.2 says it doesn’t know things instead of hallucinating. At one point it told me its last estimate was wrong and that it wouldn’t give a number until it had measured the change. Very impressive!”

CompanyCognitionAuthorAdhyyan Sekhsaria, Member of Technical Staff

Cybersecurity

Claude Mythos 5.2 is our most capable model for cybersecurity engineering. It catches more vulnerabilities than our previous models, and is better at reproducing and fixing them (note that Claude Fable 5.2’s safeguards will block most cybersecurity-related tasks). A security company found that one in six confirmed bugs found by Claude Mythos 5.2 were missed by all of the other frontier models they tested. On our internal evaluation, CVE Rediscovery,¹ Claude Mythos 5.2 found 10.2% more vulnerabilities than Claude Opus 5.5 and 14.4% more than Claude Mythos 5.1. Claude Mythos 5.2 also shows better judgement and calibration—one tester found that previous models had overstated the seriousness of about one in twenty findings, whereas Claude Mythos 5.2 hadn’t overstated any.²

CVE RediscoveryAccuracy vs. cost

• Mythos 5.2
• Opus 5.5
• Mythos 5.1

5560657075808590950Advisories found (%)0.100.200.5012510Cost per attempt (USD, log scale)LowMedHighXhighMax

CVE Rediscovery is an internal cybersecurity evaluation that assesses how many published CVEs (Common Vulnerabilities and Exposures) a model can independently find when given a snapshot of the project before a CVE was fixed. Our October 2026 run against 1,829 CVEs showed Claude Mythos 5.2 outperformed Claude Mythos 5.1 and Claude Opus 5.5 at every effort level.

Claude Mythos 5.2’s cyber capabilities are accessible through our Cyber Verification Program, which we’re continuing to expand and improve. Claude Enterprise customers (even if they aren’t members of the Cyber Verification Program) can access Claude Security, which scans codebases for vulnerabilities using our most capable models—which now includes Claude Mythos 5.2. A recent update to Claude Security helps it watch repositories as code is updated, track any findings to ensure they’re fixed, learn from additional context, and connect to a user’s tools through an API and MCP.

RubrikMozillaRubrikMozillaQuote

“We have worked closely with Anthropic’s new model, Claude Mythos 5.2, to do a next round of high-fidelity threat discovery and elimination within our own systems for potential issues we did not uncover with earlier Mythos models. The output is more readable and empowered our team to spot-check faster and more easily. We built our Code Guardian harness with Mythos 5.1, and we’d switch to Mythos 5.2 to give our customers more powerful cyber capabilities.”

CompanyRubrikAuthorArvind Nithrakashyap, Co-Founder and Chief Technology OfficerQuote

“Claude Mythos 5.2 is astonishingly good. We’ve worked closely with the Anthropic team all year to harden Firefox with successive generations of Claude models, and each round has left fewer, subtler bugs to find. Even so, Mythos 5.2 roughly doubled the vulnerability discovery metrics of Mythos 5.1 across respective scanning campaigns.”

CompanyMozillaAuthorBrian Grinstead, Distinguished Engineer

Science and engineering

Important tasks in science, engineering, and mathematics increasingly depend on coding: analyzing data, modeling molecules, designing parts, and controlling instruments are all done through software. Below, we discuss how Claude Fable 5.2 and Claude Mythos 5.2 have helped researchers make progress on difficult problems in these areas.

Theoretical physics

Claude Fable 5.2 solved two open problems in theoretical physics. Both concern black holes, and the effort to unite the two linchpin theories of modern physics, relativity, and quantum mechanics.

Both of Claude’s results rely on holographic duality, the idea that certain universes containing gravity can be precisely described by a quantum theory containing no gravity at all. In the first problem, physicists in 1983 had shown that as these universes heat up, black holes form abruptly, like water boiling; but their method worked only in large universes that were big enough for relativity to apply. A calculation to determine whether the same was true in very small universes, where only the quantum description can be used, was laid out in 2003 but was too laborious to complete. Fable 5.2, working largely autonomously, completed it and found the transition is abrupt there, too. In the second problem, Fable 5.2 found the exact limit of a technique physicists have used for more than 15 years to rule out candidate theories of quantum gravity in a simplified universe. Read more about these results in their respective preprints, here and here.

Protein folding

Adeno-associated viruses are small, harmless viruses whose protein shell, the capsid, can be used to deliver gene therapies. The protein structure of the capsid is important—for example, it affects whether the therapy will be blocked by a patient’s immune system—but it’s also highly complex. Existing protein-folding models like AlphaFold can predict individual proteins, but a capsid is built from dozens of them. Building on two open-source protein-folding models, Claude Mythos 5.2 found a novel way to accurately predict a whole capsid’s structure. It wrote code that ran these models on overlapping groups of protein chains, and used the information on the known symmetries of the capsid to merge them together.

Claude’s method worked across several capsids, both natural and engineered (in our tests, it accurately modeled 28 of the 34 it attempted)³. Three computational biologists shared the following reflections on how Claude’s skills could be used to speed up their research into gene therapies:

Dyno TherapeuticsColumbia UniversityBiohubDyno TherapeuticsColumbia UniversityBiohubQuote

“This acceleration could save meaningful time and cost from our capsid folding pipeline, enabling us to deploy compute more efficiently across many capsid properties that are needed for a genetic medicine. We are excited to help Anthropic push the reasoning capabilities further to reduce the cost of designing capsids overall.”

CompanyDyno TherapeuticsAuthorSam Sinai, Cofounder and Head of AIQuote

“Large protein assemblies, including capsids, remain an open problem for protein modeling, due to the complexity of accurately predicting and positioning the structures of dozens of individual macromolecules, not to mention the inordinately high computational costs involved. Claude’s approach cleverly tackles this problem for symmetric assemblies, making it possible for individual scientists to model systems on personal workstations that would otherwise require dedicated clusters.”

CompanyColumbia UniversityAuthorMohammed AlQuraishi, Assistant Professor of Systems Biology and Computer ScienceQuote

“Biohub’s goal in open-sourcing ESMFold2 was to enable the scientific community to build on the biology it has learned. We are excited to see Claude prototype a new sampling procedure that enables predictions of large, symmetric complexes such as AAV capsids—a protein used to deliver gene therapies. We will continue to incorporate such contributions from the community that extend our tools to important application domains.”

CompanyBiohubAuthorZeming Lin, Principal Research Scientist

Mathematics

Mathematical formalization. Formalization is the process of translating a mathematical proof into a language that a computer can check step by step. It can help human mathematicians gain confidence in results without having to check them by hand (this matters even more when AI models generate results that are hard for humans to verify). Because formalization is such a laborious process, if AI could assist with it, mathematicians could spend more time exploring novel ideas.

Our first formalization of Fermat’s Last Theorem, performed by an unreleased research version of Claude, took 11 days. It relied on a purpose-built harness along with guidance from a human researcher. By contrast, a team of 1,000 Fable 5.2 agents fully formalized the theorem in just 30 hours, with minimal human intervention. Fable 5.2 started from minimal prompts, built its own ways to coordinate, and used no mathematical references, only a list of intermediate theorem statements to score progress. It needed one-third of the action tokens as our first formalization and produced a proof seven times shorter. As can be seen in the figure below, Opus 5.5 (with the same 1,000-agent setup) completed only 26% of the formalization in 18 hours, and barely made any more progress after 30.