Claude Mythos 5.1
A version of Anthropic's Claude series of large language models, offering reasoning capabilities through API services.
Core Facts
Timeline (last 90 days)
The two model services Claude Mythos 5.1 and Claude Fable 5.1 began showing an abnormal rise in request error rates at 05:40 Pacific Time (12:40 UTC).
Officials confirmed at 06:10 Pacific Time (13:10 UTC) that the issue was resolved, with the entire window from the start of impact to full recovery lasting about 30 minutes.
Incident timeline: first marked as Investigating at 09:48 UTC, updated at 09:52 UTC, moved to Monitoring at 10:50 UTC, and marked as Resolved at 11:14 UTC.
Anthropic's status page recorded an incident involving Claude Mythos 5.1 and Claude Fable 5.1, in which both models experienced intermittent increases in error rates during a similar time period.
Fable 5.1 is open to the general public and enterprises, while Mythos 5.1 is accessible only to top-tier institutions in fields such as cybersecurity and life sciences through a 'Trusted Access Program'.
Mythos 5.1 achieved a hit rate approaching 50% across 12 targets in protein design, compared to the traditional industry standard hit rate of only 10% to 15%.
Fable and Mythos share exactly the same weight files, differing only in system prompt configuration and the strictness of the safety filtering layer.
Mythos 5.1 wrote custom GPU kernels that increased the running speed of multiple deep learning models by up to 2.5x and reduced the computational cost of whole-genome analysis.
Claude Fable 5.1 and Claude Mythos 5.1 are built on exactly the same underlying model with no difference in intelligence level; the only distinction lies in the safety protection level.
When handling cybersecurity programming tasks, false-positive interventions from AI overreaction were reduced by 60%.
7 more timeline events
All Facts (16)
The core difference between Fable 5.1 and Mythos 5.1 lies in the depth of safety measures, not in the underlying capability level.
75%UnverifiedIncident timeline: first marked as Investigating at 09:48 UTC, updated at 09:52 UTC, moved to Monitoring at 10:50 UTC, and marked as Resolved at 11:14 UTC.
60%UnverifiedAnthropic's status page recorded an incident involving Claude Mythos 5.1 and Claude Fable 5.1, in which both models experienced intermittent increases in error rates during a similar time period.
60%UnverifiedMythos 5.1 achieved a hit rate approaching 50% across 12 targets in protein design, compared to the traditional industry standard hit rate of only 10% to 15%.
50%UnverifiedFable and Mythos share exactly the same weight files, differing only in system prompt configuration and the strictness of the safety filtering layer.
50%UnverifiedFable 5.1 is open to the general public and enterprises, while Mythos 5.1 is accessible only to top-tier institutions in fields such as cybersecurity and life sciences through a 'Trusted Access Program'.
50%UnverifiedClaude Fable 5.1 and Claude Mythos 5.1 are built on exactly the same underlying model with no difference in intelligence level; the only distinction lies in the safety protection level.
50%UnverifiedMythos 5.1 wrote custom GPU kernels that increased the running speed of multiple deep learning models by up to 2.5x and reduced the computational cost of whole-genome analysis.
50%UnverifiedWhen handling cybersecurity programming tasks, false-positive interventions from AI overreaction were reduced by 60%.
50%UnverifiedMythos 5.1 can design protein binders in the life sciences field that are 10 times stronger than the industry average.
50%UnverifiedMythos 5.1 makes capabilities related to chemical and biological risks accessible only to vetted professionals through a rigorous trusted-access program.
50%UnverifiedClaude Mythos 5.1 is available only to credentialed institutions, with certain restrictions removed.
50%UnverifiedClaude Fable 5.1 belongs to the Mythos tier of the Claude 5 family and shares the same underlying architecture as Claude Mythos 5.1.
50%UnverifiedThe naming of Fable and Mythos may indicate that Anthropic is evolving from simple performance-tier classification to functional scenario-based classification.
50%UnverifiedThe two model services Claude Mythos 5.1 and Claude Fable 5.1 began showing an abnormal rise in request error rates at 05:40 Pacific Time (12:40 UTC).
50%UnverifiedOfficials confirmed at 06:10 Pacific Time (13:10 UTC) that the issue was resolved, with the entire window from the start of impact to full recovery lasting about 30 minutes.
50%