5 to 5.1 Safeguards and Access — Claude Fable and Mythos

Quick Answer

Claude Fable and Claude Mythos are not a conventional fast-versus-smart model pair. Within each generation, Anthropic says they use the same underlying model but apply different safeguard and access profiles.

Fable 5 and Mythos 5 are now the previous generation. Anthropic introduced Fable 5.1 and Mythos 5.1 on September 1, 2026, so the current decision is:

Dimension	Claude Fable 5.1	Claude Mythos 5.1
Intended access	Broad, generally available	Approved organizations only
Safeguards	General-access safeguards, with vulnerability discovery allowed and some dual-use work redirected	More permissive profile within vetted cyber or life-sciences programs
Access route	Claude products, claude-fable-5-1 API, AWS, Google Cloud, and Azure, subject to entitlement	Cyber Verification or Life Sciences Verification program plus provisioning
Self-serve sign-up	Available through normal provider products and API, subject to entitlement	None
Published price	$10 input / $50 output per million tokens; $0.25 cache reads	Verify the trusted-access agreement; do not infer entitlement from Fable pricing
Data controls	Eligible customers receive transition ZDR before phased EFS	Verify the exact program and contract; the transition statement specifically names Fable 5 and 5.1
Flowith status	Verify separately	Verify separately; provider access is restricted

Choose Fable 5.1 for a new general-access evaluation when its safeguard behavior is acceptable. Pursue Mythos 5.1 only when an eligible organization has an approved cybersecurity or life-sciences use case, governance capacity, and explicit provider provisioning.

Use the Claude Fable 5 model page and Claude Mythos 5 model page for the previous-generation entity records and migration boundaries. For subscription and API checks, use the Fable 5 plans, credits, and 5.1 migration guide.

Source check: Anthropic’s Fable 5.1 and Mythos 5.1 launch and Enterprise Frontier Safeguards announcement were checked September 2, 2026. Older 5-generation sources remain lifecycle evidence rather than the current-generation answer.

The Shared Model

Anthropic used the same underlying-model pattern twice:

• Fable 5 and Mythos 5 formed the previous generation;
• Fable 5.1 and Mythos 5.1 form the current generation.

Within the 5.1 pair, the base model is the same and the practical difference comes from safeguards and access. Fable 5.1 is generally available. Mythos 5.1 is limited to trusted programs whose professional work is affected by cybersecurity or life-sciences restrictions.

Do not attribute a base-model performance gap to the Fable and Mythos names alone. A user can still observe different task completion because safeguards, program scope, tools, system instructions, effort settings, quotas, and governance differ.

The Current Fable Safeguard Profile

Anthropic says Fable 5.1 makes its cybersecurity safeguards more precise. It can now identify software vulnerabilities for defensive work, while several dual-use tasks continue to redirect to Opus models, including:

• penetration testing;
• exploit generation;
• binary-based vulnerability scanning.

Anthropic reports around 60% fewer cyber-safeguard interventions per Claude Code session than the previous Fable 5 safeguards. That is provider evaluation evidence across its test scope, not a guarantee for a specific repository or security workflow.

For biology, Anthropic says the latest safeguards used by both Fable 5.1 and Fable 5 fire 85% less often on benign elementary biology and medical requests than the safeguards that launched with Fable 5. Research-and-development work can still be directed to Opus models.

This matters in evaluation:

• a permitted vulnerability-discovery task can still differ from exploit development;
• benign work can produce false positives;
• agent state and tool use need fallback testing;
• repeated blocked prompts are not evidence that Mythos should be self-provisioned;
• product logs should record the model or fallback shown by the provider when available.

The Current Mythos Trusted-Access Profile

Mythos 5.1 is designed for vetted people and organizations whose work is affected by Fable’s cybersecurity or life-sciences restrictions.

Anthropic describes two current program paths:

• Cyber Verification Program: Anthropic says the existing program will include Mythos-class access for vetted defensive-security work in the near future.
• Life Sciences Verification Program: Anthropic says it has enrolled initial participants and plans to expand access for professional research and development.

Anthropic currently limits Mythos 5.1 to a set of US organizations while coordinating expansion to broader domestic and international partners. Claude Security also uses Mythos 5.1, but that product fact does not give a separate customer direct model access.

The phrase “more permissive safeguards” does not mean unrestricted use. Access remains bounded by:

• organization approval;
• program purpose;
• provider and government coordination where applicable;
• acceptable-use terms;
• monitoring and review;
• security and incident-response controls;
• human oversight;
• retention requirements.

Access Decision Tree

Choose Fable 5.1 When

• the user or organization needs the current generally available Fable generation;
• normal Claude products, API, or supported cloud access fits;
• conservative fallback behavior can be tested and accepted;
• the current plan or usage-credit cost is approved;
• covered-model retention is compatible with the data policy.

Consider Mythos 5.1 Only When

• the person or organization is eligible for the Cyber Verification or Life Sciences Verification program;
• the use case is within the approved professional domain;
• provider approval and provisioning are complete;
• governance, logging, human review, and incident response are documented;
• the organization accepts the covered-model data terms;
• no public or lower-risk model satisfies the requirement.

Choose Another Model When

• Fable’s safeguards cause unacceptable false positives;
• Fable’s price or latency does not fit;
• the workload does not need Fable-level capability;
• Mythos access is unavailable or inappropriate;
• retention conflicts with organizational requirements.

Claude Opus 5, Sonnet 5, Fable 5, or another supported model may provide a better compatibility, price, latency, governance, or access fit.

API and Provider Access

Fable

The previous first-party API ID is:

claude-fable-5

The current generally available first-party API ID is:

claude-fable-5-1

Anthropic also lists Fable 5.1 across AWS, Google Cloud, and Microsoft Azure. Cloud availability still depends on the current region, platform ID, quota, and contract.

Mythos

The previous-generation restricted ID was:

claude-mythos-5

The ID is not a self-serve credential and does not grant current 5.1 access. Approved customers should use the exact endpoint and terms provisioned through the relevant trusted program. The September 1 launch page does not provide a public self-serve Mythos 5.1 API path.

Do not:

• probe restricted IDs from an unapproved account;
• infer access from a documentation table;
• copy Fable credentials or entitlements;
• market either Mythos generation as a paid consumer upgrade;
• imply Flowith access from a provider model ID.

Price Is Not the Access Gate

Anthropic published $10 per million input tokens and $50 per million output tokens for Fable 5 and keeps those base rates for Fable 5.1. Fable 5.1 cache reads cost $0.25 per million tokens.

That does not make Mythos 5.1 purchasable on demand or establish its contract price. Trusted-program approval and scope are separate from Fable consumption pricing.

For Fable, price also appears through different billing surfaces:

• standard API consumption;
• usage credits in Claude products;
• cloud platform billing;
• enterprise contracts.

Verify prompt caching, cache creation, cache reads, batch pricing, effort, tools, long outputs, retries, and platform markup before estimating a workload. Anthropic’s 25% typical and up-to-45% agentic savings are provider estimates, not guarantees for one workload.

Enterprise Frontier Safeguards and Retention

Anthropic introduced Enterprise Frontier Safeguards to combine zero-data-retention privacy with misuse detection that can correlate activity across time and accounts. It says eligible Fable 5 and Fable 5.1 customers receive zero data retention until EFS is ready.

Under EFS:

• monitoring activity data can be stored in the customer’s AWS, Azure, or Google Cloud account;
• customer-managed encryption keys, access policies, and audit logs can govern the data;
• automated systems analyze a rolling traffic window and send serious-misuse flags to the customer;
• the customer’s authorized reviewers decide what to do next;
• no Anthropic human review is required;
• customer-owned storage, customer-managed keys, and automated review are separate opt-in controls;
• Anthropic does not charge for EFS, but the cloud provider can charge for storage operations and egress.

EFS is rolling out in phases and does not erase the need to verify the exact product, eligibility, contract, storage, review, and retention terms. Anthropic’s transition statement specifically names eligible Fable 5 and Fable 5.1 customers; do not silently extend it to a Mythos 5 or Mythos 5.1 agreement.

Flowith Availability Is Independent

Neither general Fable availability nor approved Mythos access establishes:

• a Flowith mapping;
• plan entitlement;
• model routing;
• provider fallback visibility;
• context and output limits;
• pricing;
• retention or data handling.

Use the live Flowith workspace and first-party Flowith documentation. Anthropic’s trusted programs, Claude products, cloud platforms, and Flowith are distinct surfaces.

Evaluation Checklist

Test Fable 5.1 before pursuing restricted access:

1. representative allowed prompts;
2. cyber, biology, or chemistry-adjacent benign prompts likely to trigger safeguards;
3. user-visible fallback notification;
4. model identity and audit logging;
5. tool execution after fallback;
6. latency and output consistency;
7. plan or API cost;
8. retention and data classification;
9. human review of high-impact outputs;
10. a lower-risk fallback model.

If Fable fails the acceptance bar, document the exact blocked capability and business need. That evidence may support a provider conversation; it does not create Mythos eligibility.

Frequently Asked Questions

Is Mythos 5 more capable than Fable 5?

Anthropic described one underlying model for the 5 pair and now describes one underlying model for the 5.1 pair. Mythos can expose professional capabilities that Fable’s safeguards redirect, so the practical difference is safeguard-mediated access rather than a published base-model capability tier.

Can I pay extra to unlock Mythos 5?

No self-serve path is documented for Mythos 5 or Mythos 5.1. A current applicant must qualify for the Cyber Verification or Life Sciences Verification program and receive provisioning.

Does Fable 5 always answer with Fable 5?

Not necessarily. Historical Fable 5 guidance documented biology fallback to Opus 5. The current Fable 5.1 launch says professional life-sciences work and several dual-use cyber tasks can be redirected to Opus models. Test the exact product behavior rather than assuming one universal fallback.

Does Mythos 5 have different pricing?

Anthropic published the same base token pricing for the previous 5 pair. Do not infer Mythos 5.1 pricing or entitlement from Fable 5.1’s public rates; the approved program and governing contract control.

Should a new project evaluate the 5 or 5.1 generation?

Evaluate Fable 5.1 first for general access and pursue Mythos 5.1 only through a relevant trusted program. Keep the 5 generation where compatibility, an existing production dependency, or a governing contract requires it.

Does EFS make Mythos 5.1 zero-data-retention by default?

No such universal statement is supported. EFS is phased and opt-in, and Anthropic’s transition ZDR statement specifically names eligible Fable 5 and Fable 5.1 customers. Verify the exact Mythos 5.1 program agreement.

Official Sources

• Anthropic: Introducing Claude Fable 5.1 and Claude Mythos 5.1
• Anthropic: Enterprise Frontier Safeguards
• Anthropic: Claude Fable 5 and Claude Mythos 5
• Anthropic: Redeploying Claude Fable 5
• Anthropic: Improving Fable 5 safeguards
• Claude Platform model overview
• Claude Fable 5 on your plan
• Data retention practices for covered models