Anthropic introduced Claude Fable 5.1 and Claude Mythos 5.1. Fable 5.1 is generally available, while Mythos 5.1 stays limited to trusted access programs. The split puts cost, access, and safeguards into different lanes for enterprise teams and technical users.
Claude Fable 5.1 pricing
Anthropic says Claude Fable 5.1 will cost an estimated 25% less than Claude Fable 5 for typical workloads billed by token. That is the number buyers should use when they compare model runs inside Claude Code, Claude Cowork, or Claude.ai.
Anthropic also says savings on highly agentic work can often be much larger and can reach approximately 45%. That gives teams a narrower bill for long-running tasks, but only if their workload fits the pattern Anthropic is describing.
Claude Mythos 5.1 access
Anthropic says Claude Mythos 5.1 remains in trusted access programs because it carries stronger controls around advanced biology capabilities. The company has also established an access program in partnership with the US government for that work.
Anthropic expects to open enrollment for scientists soon. For anyone building around biology-related use, that means access is moving through a controlled program rather than landing as a broad public release.
Enterprise Frontier Safeguards
Anthropic says Enterprise Frontier Safeguards give customers complete privacy while staying state-of-the-art at preventing adversarial use. The system stores data in cloud infrastructure controlled entirely by the customer, not Anthropic, which is the practical difference between a standard hosted setup and customer-controlled storage.
Anthropic says those safeguards will reach enterprise customers in phases beginning later this fall. Until then, eligible customers can use Fable 5.1 with zero data retention.
Claude Code and cyber use
Anthropic says its newest cybersecurity safeguards cut false positives by 60% from before. That should reduce the amount of harmless activity that gets flagged during security work, which is the main operational gain for teams trying to separate noise from real risk.
The company also says Fable 5.1 can be used to discover software vulnerabilities but not develop exploits. Anthropic is drawing a line between finding weaknesses and turning them into weaponized code, and that line is the part security teams will want to test in practice.
Anthropic says Fable 5.1 and Claude Mythos 5.1 are the same model with different levels of safeguards. The company says the gap between them reflects the tasks on which earlier, less precise cyber safeguards intervened, so the naming split is less about two engines than about two policy envelopes around one base model.
Anthropic says Fable 5.1 defaults to High effort in Claude Code and to Medium in Claude Cowork and on Claude.ai. For users, that means the same model can behave differently depending on where it runs, which can shift latency, cost, and how aggressively it reasons through a task.
Millennium’s testing gives the launch a concrete stress case. Anthropic says Fable 5.1 found the cause of a rare crash on Millennium’s internal systems after several years of trying by its engineers and other models, which is the kind of claim enterprise buyers will read as a proof point rather than a guarantee.
The unresolved issue is how quickly Enterprise Frontier Safeguards will reach each enterprise customer after the phased rollout begins later this fall.
