Base model
Same as Fable 5.1
Anthropic differentiates the pair through safeguard and access profiles, not a public fast-versus-smart tier.
Anthropic trusted-access record · checked September 2, 2026
Answer first
It shares Fable 5.1's underlying model but uses different safeguards for approved cyber and life-sciences work. Program eligibility, provisioning, endpoint, data terms, price, and human oversight are part of the model contract.
Base model
Anthropic differentiates the pair through safeguard and access profiles, not a public fast-versus-smart tier.
Access
No consumer upgrade or public self-serve API path is documented in the September 1 launch.
Cyber path
The Cyber Verification Program will include Mythos-class access for vetted defensive-security work.
Life-sciences path
Initial participants are enrolled; broader professional access is planned but must not be assumed.
Use Fable 5.1 or another generally available model. A public Fable ID, paid plan, or cloud account does not unlock Mythos.
Use the Cyber Verification Program route and wait for explicit Mythos-class approval, provisioning, program scope, and endpoint instructions.
Use the Life Sciences Verification Program route. Anthropic says its safeguards support professional R&D while other safeguards remain in place.
Anthropic says access is currently limited to a set of US organizations while it coordinates broader domestic and international expansion. Verify current enrollment evidence.
Claude Security uses Mythos 5.1 to scan codebases and suggest patches for human review. That is a product-surface fact, not a separate entitlement to the raw model. Compare with the generally available Fable 5.1 record and the 5-to-5.1 safeguard guide.
Shared model, safeguard differences, verified programs, current US access, planned expansion, Claude Security, and public Fable boundaries.
EFS architecture and the exact Fable transition statement used to avoid extending unsupported data claims to Mythos.
Preserve the previous ID, program history, specs, and migration boundary.
Check storage, keys, monitoring, reviewers, cost, tests, and rollback without assuming coverage.
Use the current generally available model when trusted access is unavailable or unnecessary.