How to Get Access to Claude Mythos 5.2 (November 2026)
The short answer
Apply to the Cyber Verification Program (CVP) in Anthropic’s Verification Portal, pick the lowest tier that covers your work, and attest the security controls that tier requires. Claude Mythos 5.2 launched on November 9, 2026 alongside Claude Fable 5.2; they are the same model, and Mythos has fewer cyber and biology safeguards. As of November 16, 2026 it is available only to vetted organisations and individuals through two programs:
- Cyber Verification Program — for security work. Apply at
portal.anthropic.com/programs. - Life Sciences Verification Program — for biology and life sciences research.
Pricing: $8 input / $40 output per million tokens, fast mode $16 / $80 (see our current API prices).
Step 1 — Check whether you need Mythos at all
Anthropic’s own guidance: you do not need CVP for everyday secure development — reviewing code, patching known issues, triaging alerts — on generally available models. You need it if Claude blocks legitimate security work such as validating exploits or running authorised penetration tests. If you only want Fable 5.2’s capability, read Fable 5.2 vs Mythos 5.2 vs Opus 5.5 first.
Step 2 — Pick your tier
| Defense Access | Red Team Access | Specialized Access | |
|---|---|---|---|
| Work allowed | Security operations, incident response, malware analysis, detection engineering, vulnerability triage and validation | Authorised pen testing, red teaming, adversary emulation, offensive tool and exploit development | Authorised testing of critical infrastructure; research into the highest-severity techniques |
| Who can apply | Organisations and individuals | Organisations only | Organisations only |
| Plans | Pro, Max, Ultra, Team, Enterprise, API | Team, Enterprise, API | Team, Enterprise, API |
| Approved users | No default limit | Up to 75 | Up to 25 |
| Controls to attest | MFA; API keys replaced at least every 7 days | Phishing-resistant MFA, 7-day key rotation, company-domain accounts, organisation-controlled devices, outbound network controls, background checks | All Red Team controls plus SSO and managed devices with malware protection; reviewed with the US government |
Every tier includes Claude Mythos 5.2, Mythos 5.1, Opus 5.5 and Sonnet 5.5; the tier changes how much the safeguards intervene. Anthropic notes that operating critical infrastructure does not by itself qualify for Specialized Access — Defense Access suits most operators. If you apply for a higher tier and qualify only for a lower one, you get the lower tier while the rest is reviewed.
Step 3 — Link your account, then apply
- Decide how you will call the model. Directly through Anthropic (Claude apps, Claude Code, Claude Console), or through Claude Platform on AWS, Claude Platform on Google Cloud, Google Cloud Agent Platform or Microsoft Foundry. For cloud accounts, link the account or project in the Verification Portal before you apply (on Foundry, deploy a Claude model first).
- Verify as a business or an individual. Choose carefully: the verification type affects which tiers you can get.
- Describe your security work, attest the controls for your tier, and accept the program terms. Usage Policy still applies in full.
- Wait for the email. Anthropic replies with a decision or a request for more information; check status under Programs → Cyber Verification Program in the portal.
Step 4 — Turn access on
After approval, an admin turns the tier on: in the Verification Portal under “Access on linked accounts” for cloud and third-party platforms, by workspace in Claude Console, or by role in Claude Enterprise. On cloud partners, Mythos access starts about five business days after approval; customers with a committed contract complete a short form and accept a private offer in their cloud marketplace. Amazon Bedrock requires Enterprise Frontier Safeguards (EFS), which is not yet generally available.
What else to know
- Data retention is required. Mythos models need 30-day retention for safety monitoring by default. Organisations that already have a zero-data-retention exemption for Fable or Mythos models keep ZDR for internal CVP use; EFS will later allow storage in your own cloud.
- Internal use only by default. Offering Mythos-powered security work to customers needs a separate approval per service: managed services (your staff deliver reviewed results) are accepted now; products where customers prompt the model are at the “express interest” stage.
- Third-party platforms powered by Claude can offer Defense and Red Team Access; ask the platform to link your account.
For how access worked under the previous version, see what is Claude Mythos 5.1 trusted access.
Last verified: November 16, 2026.