CLI Coming Soon
    Back to catalog

    Claude Mythos 5.2 vs Fable 5.2: which to use

    Published · Last updated

    Abstract cover: two identical dark panels side by side, the right one marked with an orange bar

    Short answer

    Claude Fable 5.2 and Claude Mythos 5.2 are the same underlying model at the same price, $8 input and $40 output per million tokens. Fable carries cyber and biology safeguards and is open to all users. Mythos is for vetted organizations in Anthropic's verification programs.

    Anthropic released both on November 9, 2026 (announcement). Every cell below quotes and links one of four Anthropic pages: the announcement, the Claude Mythos page, the models overview and the pricing docs. All were checked on 2026-11-11. Where none of them states a fact, the cell reads "Anthropic does not say".

    Side by side

    Claude Mythos 5.2 and Claude Fable 5.2, with quoted sources
    ItemClaude Fable 5.2Claude Mythos 5.2
    Underlying modelThe same model. Anthropic gives one statement for both.Announcement"Claude Fable and Claude Mythos are the same underlying model."
    Availability and accessAll users, on all platforms.Announcement"Fable has strong cybersecurity and biology safeguards to guard against misuse, which allows us to make it available to all users.""Claude Fable 5.2 is available today on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure."Vetted organizations, through Anthropic's Cyber and Life Sciences Verification Programs. Claude Enterprise customers can also use it inside Claude Security. The pricing docs label it "limited availability".Mythos page"Claude Mythos 5.2 is available to vetted cyberdefenders and life scientists through our trusted access programs. Life sciences researchers can apply for advanced capabilities through the Life Sciences Verification Program."Announcement"Cybersecurity professionals can apply for access to Claude Mythos 5.2 through the Cyber Verification Program. Claude Enterprise customers can also use Claude Security, powered by our full family of models, including Claude Mythos 5.2."
    Model IDclaude-fable-5-2 on the Claude API, Claude Platform on AWS, Google Cloud and Microsoft Foundry. anthropic.claude-fable-5-2 on Amazon Bedrock.Models overviewAnthropic does not sayThe models overview has no Mythos 5.2 column. The announcement and the Mythos page give no ID.
    Price per million tokens$8 input, $40 output. Anthropic gives one price for both models.Announcement"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."Pricing docs: separate rows for Claude Fable 5.2 and Claude Mythos 5.2, each "$8 / MTok" base input and "$40 / MTok" output.Mythos page"Pricing for Claude Mythos 5.2 starts at $8 per million input tokens and $40 per million output tokens."
    Batch and cache readsBatch: $4 input, $20 output per million tokens. Cache hits: $0.20 per million tokens. Same rows for both models.Pricing docs"On Claude Fable 5.2, Claude Mythos 5.2, Claude Fable 5.1, and Claude Mythos 5.1, a cache hit costs 2.5 percent of the standard input price ($0.20 USD per million tokens on Claude Fable 5.2 and Claude Mythos 5.2 ..."
    Fast mode$16 input, $80 output per million tokens. Up to 2.5x faster token generation.Announcement"Fast mode is available for $16 per million input tokens and $80 per million output tokens.""Fast mode for Claude Fable 5.2 and Claude Mythos 5.2 is available on Claude Code, the Claude Platform, Claude Models on Google Cloud (Anthropic managed and Google managed), and Microsoft Foundry (Anthropic hosted)."$16 input, $80 output per million tokens. Verification Program members only.Announcement"Note that access to Claude Mythos fast mode is available only to organizations who are part of our Verification Programs."Mythos page"It costs $16 per million input tokens and $80 per million output tokens."
    Context window and max output1M token context window. 128K max output tokens.Models overviewAnthropic does not sayThe models overview lists these limits for Fable 5.2 only.
    Intended useDemanding reasoning and long-horizon agentic work.Models overview"For demanding reasoning and long-horizon agentic work"Announcement"Fable and Mythos are best for the most ambitious, open-ended technical work."Cybersecurity and life sciences research.Mythos page"Claude Mythos 5.2 is our most capable model for cybersecurity and life sciences research."Announcement"Claude Mythos 5.2 is our most capable model for cybersecurity engineering."
    Published benchmarks11 rows in Anthropic's table. See the benchmark table below.4 rows in Anthropic's table, marked "Mythos 5.2". See the benchmark table below.
    SafeguardsClassifiers block high-risk cyber work and most research biology. Blocked queries get an answer from an older model.Announcement"Fable 5.2 comes with safety classifiers that block high-risk cyber activities, including penetration testing, exploit generation, and binary-based vulnerability, as well as most tasks in research biology. When a query is blocked, users will instead receive a response from Opus 4.8 for cybersecurity and Opus 5 for biology."Without certain of Fable's cyber and biology safeguards. Access is limited to vetted organizations.Announcement"Claude Mythos 5.2 (the same underlying model as Fable 5.2 without certain cybersecurity or biology safeguards)"Mythos page"Because Mythos 5.2 is highly capable for cybersecurity and biology research, it could be used both for good and for harm. We currently provide access to a set of vetted organizations through our verification programs."
    Data retention30 days for safety monitoring, unless the customer is eligible for zero data retention.Announcement"Eligible customers can continue using Claude Fable 5.2 and Claude Mythos 5.2 with zero data retention before implementing safety monitoring with EFS. For other customers, we have 30-day retention for safety monitoring."30 days by default. Cyber Verification Program customers who already have zero data retention can keep it.Mythos page"Using Mythos models requires 30-day data retention for safety monitoring by default.""Until EFS is available, eligible Cyber Verification Program customers who already have ZDR can keep it for Mythos 5.2."
    Rows that span both columns are cases where Anthropic gives one statement or one price for both models. Sources: Anthropic, "Introducing Claude Fable 5.2 and Claude Mythos 5.2" (November 9, 2026), the Claude Mythos page, and the Claude Platform models overview and pricing docs. Checked 2026-11-11.

    Published benchmarks

    The announcement and the Mythos page show the same benchmark table. In it, Anthropic reports most rows for Fable 5.2 and four rows for Mythos 5.2. The Mythos page explains the split: "Results below are for Fable 5.2 unless noted; where Mythos 5.2 results are shown, the gap reflects tasks where Fable 5.2's safeguards intervene." (Mythos page)

    Benchmarks Anthropic published for Claude Fable 5.2 and Claude Mythos 5.2
    BenchmarkFable 5.2Mythos 5.2
    FrontierCode 2.0Agentic coding49.1% (Xhigh effort)1Anthropic does not say
    FrontierSWE v2Agentic coding69.9%Anthropic does not say
    AutomationBenchBusiness workflows51.4%Anthropic does not say
    Terminal-Bench-Science 0.1Agentic scientific research74.6%75.3%
    CyScenarioBench †CybersecurityAnthropic does not say76.3%
    CVE RediscoveryCybersecurityAnthropic does not say89.6%
    De novo binder design (24 hr)Protein designAnthropic does not say85.12
    WANDR (relaxed score)Agentic search82.1%Anthropic does not say
    OSWorld 2.1 (offline set, partial)Computer use89.3%Anthropic does not say
    RoboDojo Sim ‡Robotics64.9%Anthropic does not say
    AA-Briefcase v1.1Knowledge work66.1% rubric, 1785 EloAnthropic does not say
    Remote Labor Index (RLI)Knowledge work41.3% (High effort)Anthropic does not say
    Humanity's Last Exam (with tools)Multidisciplinary reasoning74.2%Anthropic does not say
    Chartography (no tools)Visual chart recognition73.4%Anthropic does not say
    Anthropic's notes under the table (Announcement): "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."1. Effort as labelled in Anthropic's table. The announcement adds: "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".2. The announcement shows "85.1" with no unit. The Mythos page shows "85.1%".Methodology is in the Claude Fable 5.2 and Mythos 5.2 System Card.

    Which to use

    This section only restates Anthropic's own guidance, quoted from the pages above.

    • Not sure which model. The models overview says: "If you're unsure which model to use, start with Claude Opus 5.5 for most workloads. Use Claude Fable 5.2 for demanding reasoning and long-horizon agentic work, or when your evals on Claude Opus 5.5 at higher effort still fall short." (Models overview)
    • Most coding and knowledge tasks. The announcement says: "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". It names the direction as "upward to Fable or Mythos, or downward to Claude Sonnet 5.5 or Claude Haiku 5.5." (Announcement)
    • Ambitious technical work. "Fable and Mythos are best for the most ambitious, open-ended technical work." (Announcement)
    • Fable 5.2 with smaller models. "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)." (Announcement)
    • Cybersecurity work. "Claude Mythos 5.2 is our most capable model for cybersecurity engineering." The same paragraph notes "that Claude Fable 5.2's safeguards will block most cybersecurity-related tasks". Access is through the Cyber Verification Program (Announcement).
    • Codebase scanning on Claude Enterprise. "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". Anthropic adds that this "now includes Claude Mythos 5.2" (Announcement, Claude Security).
    • Life sciences research. "Life sciences researchers can apply for advanced capabilities through the Life Sciences Verification Program." (Mythos page, Life Sciences Verification Program)

    Cost

    Anthropic's pricing docs list the same token rates for both models (Pricing docs). To estimate a task, looot's Fable 5.2 vs Opus 5.5 cost per task calculator uses Fable 5.2's $8 / $40 rates.

    For Claude Code setup, model switching and cost with Fable 5.2, see Claude Code with Claude Fable 5.2: setup and cost. For Claude Mythos Preview's earlier work in Project Glasswing, see the Project Glasswing explainer.

    Anthropic's prompting page covers both models. Its sections on effort, approvals, subagents and refusals, with CLAUDE.md before and after: Prompting Claude Fable 5.2: Anthropic's rules.

    Sources

    All fetched and checked 2026-11-11.

    Related guides