At a glance
Fable access: Anthropic says: “Claude Fable 5.2 is available today on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure.”
Mythos access: Anthropic says: “Cybersecurity professionals can apply for access to Claude Mythos 5.2 through the Cyber Verification Program.” It adds: “Life sciences researchers can also apply for access through the Life Sciences Verification Program.”
Pricing: Anthropic says Fable 5.2 and Mythos 5.2 are “available at a lower price: $8 per million input tokens and $40 per million output tokens.” Optional fast mode is “$16 per million input tokens and $80 per million output tokens.”
Official announcement: Claude Fable 5.2 and Mythos 5.2
What shipped
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Nov 09 2026. Anthropic says the two are the same underlying model. Fable carries strong cybersecurity and biology safeguards and is available to all users. Organizations pursuing beneficial applications in cybersecurity or biomedical research can apply for Mythos through its Cyber and Life Sciences Verification Programs. Both cost $8 per million input tokens and $40 per million output tokens, which Anthropic describes as a lower price.
Key capabilities
- Coding with Fable 5.2. Anthropic says a team of agents running an earlier version of Fable 5.2 migrated a critical production service from Python to Rust. In an experiment, the model coordinated over 1,000 subagents for eight days and produced a working operating system that boots on a laptop.
- Fable 5.2 as Advisor or Orchestrator. Anthropic says Fable 5.2 can serve as an Advisor, which Haiku, Sonnet or Opus models consult on harder problems, or as an Orchestrator that delegates simpler tasks to smaller models.
- Cybersecurity engineering with Mythos 5.2. Anthropic calls Mythos 5.2 its most capable model for this work. On its internal CVE Rediscovery evaluation, Mythos 5.2 found 10.2% more vulnerabilities than Claude Opus 5.5 and 14.4% more than Claude Mythos 5.1. Fable 5.2's safeguards block most cybersecurity tasks.
- Field reports, with a caveat. Anthropic says a security company found that one in six confirmed bugs found by Mythos 5.2 were missed by all of the other frontier models it tested. One tester found previous models overstated the seriousness of about 1 in 20 findings, while Mythos 5.2 overstated none. Anthropic's footnote says those counts are raw findings from one tester's audit, and that on its own evaluations vulnerability identification is roughly on par with Opus 5.5.
- Speed. Anthropic says the models generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1. An optional fast mode speeds token generation by up to 2.5 times.
- Life sciences with Mythos 5.2. Mythos 5.2 found what Anthropic calls a novel way to predict the structure of a whole adeno-associated virus capsid. In Anthropic's tests it accurately modeled 28 of the 34 capsids it attempted.
- Alignment testing. Anthropic says the model performed as well as, or better than, its recent models on most alignment measures. In a sandbox test of over 1,800 scenarios, Mythos 5.2 used credentials or files placed outside its sandbox in 3.1% of scenarios, against 3.3% for Opus 5.5 and 21% for Mythos 5.1. On other measures it scored slightly worse than Opus 5.5.
- Limits set on purpose. Anthropic says it limited advances during training in the research capabilities that could most directly accelerate AI development. In areas such as accelerator kernel optimization, both models perform at about the level of Opus 5.5.
Benchmarks and comparisons
Anthropic-reported results from the official announcement, accessed Nov 09 2026. These are publisher-reported measurements; test settings can differ across models. Anthropic heads its first results column Fable 5.2; cells marked Mythos 5.2 are Mythos 5.2 results.
Benchmark | Fable 5.2 | Fable 5.1 | Opus 5.5 | GPT-6 Astra | Gemini 4 Carbon |
|---|---|---|---|---|---|
Agentic coding FrontierCode 2.0 | 49.1% Xhigh effort | 32.4% Xhigh effort | 45.5% | 42.8% | 41.1% Medium effort |
Agentic coding FrontierSWE v2 | 69.9% | 56.3% | 62.3% | 65.5% | 61.2% |
Business workflows AutomationBench | 51.4% | 31.4% | 42.5% | 41.4% | 53.9% |
Agentic scientific research Terminal-Bench-Science 0.1 | 74.6% 75.3% (Mythos 5.2) | 52.6% | 58.7% | 68.1% | 64.8% |
Cybersecurity CyScenarioBench † | 76.3% Mythos 5.2 | 61.7% Mythos 5.1 | 67.6% without safeguards | 61% GPT-6 Cyber | — |
Cybersecurity CVE Rediscovery | 89.6% Mythos 5.2 | 78.3% Mythos 5.1 | 81.3% without safeguards | — | — |
Protein design De novo binder design (24 hr) | 85.1% Mythos 5.2 | 79.1% Mythos 5.1 | 82.6% without safeguards | — | — |
Agentic search WANDR (relaxed score) | 82.1% | 68.7% | 72.3% | — | — |
Computer use OSWorld 2.1 (offline set, partial) | 89.3% | 83.5% | 87.2% | 79.5% | 81.5% |
Robotics RoboDojo Sim ‡ | 64.9% | 27.6% | 39.4% | 29.8% | — |
Knowledge work AA-Briefcase v1.1 | 66.1% rubric | 61.5% rubric | 61.2% rubric | 52.0% rubric | 61.8% rubric |
Knowledge work AA-Briefcase v1.1 | 1785 Elo | 1669 Elo | 1800 Elo | 1569 Elo | 1611 Elo |
Knowledge work Remote Labor Index (RLI) | 41.3% High effort | 17.9% | 21.3% | 20.8% | — |
Multidisciplinary reasoning Humanity’s Last Exam (with tools) | 74.2% | 65.6% | 67.7% | 57.2% | — |
Visual chart recognition Chartography (no tools) | 73.4% | 44.8% | 64.4% | 71.0% | 72.5% |
Anthropic's notes: 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.
Anthropic also reports these results alongside its charts. Cost figures are Anthropic's cost per task.
- FrontierCode 2.0: at High effort, Fable 5.2 scored 45.3% against Opus 5.5's best score of 45.5%, for roughly half the cost per task.
- WANDR: at Medium effort, Fable 5.2 matched Opus 5.5's top score for about 40% of the cost per task.
- BenchCAD 2.1: at Max effort, Fable 5.2 scored 83.5, ahead of Opus 5.5 (65.3) and Fable 5.1 (44.4).
- Humanity's Last Exam: at Low effort, Fable 5.2 matched Opus 5.5's top score for less than one-tenth of the cost per task.
- Chartography: at Medium effort, Fable 5.2 scored 66.4%, ahead of Opus 5.5 at Max effort (64.4%), for less than one-tenth of the cost per task.
- CVE Rediscovery: in an October 2026 run against 1,829 CVEs, Mythos 5.2 outperformed Mythos 5.1 and Opus 5.5 at every effort level.
- Fermat's Last Theorem formalization: about 1,000 Fable 5.2 agents fully formalized the theorem in 30 hours. With the same setup, Opus 5.5 completed 26% in 18 hours and made little more progress after 30. Anthropic reports a single run per model.
Official samples and early work
- Official sample: a working operating system built by Fable 5.2
- Official sample: capsid structure work by Mythos 5.2
- Official sample: electronic devices designed by Fable 5.2 with Diode
- Official sample: robots controlled by Fable 5.2
What it means for agent work
For most operations teams, this release is a routing change: Fable 5.2 is open to everyone, and Mythos 5.2 is an access decision before it is a model decision.
Pick by what the safeguards block. Fable 5.2 is the default of the two. Mythos 5.2 is for verified cybersecurity or biomedical work that Fable's safeguards would stop.
Treat access as a routing condition. An agent that depends on Mythos 5.2 should stop and name an owner when an account lacks access, rather than quietly continuing on another model.
Expect a different model on blocked requests. Anthropic says Fable 5.2 blocks high-risk cyber activity, as well as most tasks in research biology, and that users instead receive a response from Opus 4.8 for cybersecurity and Opus 5 for biology. Check which model answered before a result moves to the next step.
Gate verified access like a privileged credential. Mythos 5.2 runs without certain safeguards, so who can send it which work is a policy question. Archon's security page lists "SSO, RBAC, approval queues, and audit logs" as its access model. For the design detail, see governed AI agents need approval queues and audit trails.
Check retention before moving sensitive work. Enterprise Frontier Safeguards, now in private beta, is compatible with zero data retention. For other customers, Anthropic has 30-day retention for safety monitoring.
Route by task, not by release. Anthropic says Fable and Mythos are best for the most ambitious, open-ended technical work, and recommends most coding and knowledge tasks start on Claude Opus 5.5. Archon is "model agnostic by design", so a release like this is a routing update rather than a rebuild; see why enterprise AI should be model agnostic.
Test on your own work. The results above are Anthropic's, and it says the external evaluators did not review its announcement. Run a sample of your own code first, and have a person confirm findings before any fix ships.
To work through access, approvals and routing for your team, Book a Consultation.
How to try it
- Claude Fable 5.2: available today on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. Developers can start with claude-fable-5-2 on the Claude Platform.
- Cybersecurity teams: apply for Mythos 5.2 through the Cyber Verification Program.
- Life sciences researchers: apply for Mythos 5.2 through the Life Sciences Verification Program.
- Claude Enterprise customers: Anthropic says they can use Claude Security, powered by models that include Mythos 5.2.
- Fast mode: available on Claude Code, the Claude Platform, Claude Models on Google Cloud and Microsoft Foundry. Mythos fast mode is limited to organizations in the Verification Programs.
Sources
- https://www.anthropic.com/claude-fable-and-mythos-5-2, accessed Nov 09 2026