ANTHROPIC · MODEL RELEASE TRACKER
Claude Mythos
Claude Mythos is a model from Anthropic. In an Anthropic announcement about expanding a partnership with the U.S. GSA, Anthropic reports that a U.S. federal security team used Claude Mythos to detect and patch a vulnerability that would have allowed takeover of a critical production system, accomplishing this in about two days.CURRENT SNAPSHOT2/5 DIMENSIONS WITH DATA
The dimensions that change the decision.
OSS Scanner is offered at no cost to participating open-source projectsAnthropic's OSS Scanner, which uses its strongest models including Claude Mythos, will be provided to opt-in open-source projects free of charge.
Not established from the available sources.
Not established from the available sources.
Not established from the available sources.
Preview access restrictedAnthropic had a Mythos preview that was restricted to participating organisations.
RELEASE TIMELINE
Published, source-backed release events only.
VERIFIABLE FACTS
Every value stays attached to a source and date.
CAPABILITIES · Claim about vulnerability-detection capabilityDEVELOPER CLAIM
Anthropic claimed that Claude Mythos "is better at finding software vulnerabilities than most security experts." This is reported as a company claim (end of April) and is not presented here as independently verified.
CAPABILITIES · Software vulnerability detection (claimed)DEVELOPER CLAIM
Anthropic claimed that Claude Mythos "is better at finding software vulnerabilities than most security experts" (reported at the end of April).
CAPABILITIES · Used to generate OSS Scanner vulnerability reportsDEVELOPER CLAIM
Anthropic’s OSS Scanner generates vulnerability reports using its strongest models, including Claude Mythos, to scan open-source projects.
CAPABILITIES · Described as one of Anthropic’s 'strongest models'DEVELOPER CLAIM
Claude Mythos is described in reporting as one of Anthropic’s “strongest models.”
PRICING · OSS Scanner is offered at no cost to participating open-source projectsDEVELOPER CLAIM
Anthropic's OSS Scanner, which uses its strongest models including Claude Mythos, will be provided to opt-in open-source projects free of charge.
LIMITATIONS · OSS Scanner outputs are fully model-generated without human review and may be incorrectDEVELOPER CLAIM
OSS Scanner’s outputs will be fully model-generated without human review or triage, which means it is possible the reports will be incorrect or invalid.
CAPABILITIES · Used as one of Anthropic's strongest models for vulnerability scanningDEVELOPER CLAIM
Anthropic says OSS Scanner's vulnerability reports are generated by its strongest models, including Claude Mythos.
AVAILABILITY · Preview access restrictedDEVELOPER CLAIM
Anthropic had a Mythos preview that was restricted to participating organisations.
LIMITATIONS · OSS Scanner reports are fully model-generated without human review and may contain errorsDEVELOPER CLAIM
Anthropic says OSS Scanner outputs — generated by models including Claude Mythos — will be fully model-generated without human review, which means reports may be incorrect or invalid.
CAPABILITIES · Used to generate OSS Scanner security reportsDEVELOPER CLAIM
Anthropic's OSS Scanner generates periodic security reports for enrolled open-source project maintainers using the company's strongest models, including Claude Mythos.
CAPABILITIES · Reports sent without human review; include reproducers, explanations, identified changes and candidate patchesDEVELOPER CLAIM
Reports generated by Anthropic's models including Claude Mythos are sent to maintainers without human review and include a self-contained reproducer, an explanation of the suspected vulnerability, bisection to identify the introducing change when possible, and a candidate patch.
CAPABILITIES · Used to detect and patch a security vulnerabilityDEVELOPER CLAIM
Anthropic reports that a U.S. federal security team used Claude Mythos to detect and patch a vulnerability that would have allowed takeover of a critical production system, accomplishing this in about two days.
WHAT CHANGED
Stored passport versions, without reconstructed history.
CAPABILITIES · Used to detect and patch a security vulnerability+ Anthropic reports that a U.S. federal security team used Claude Mythos to detect and patch a vulnerability that would have allowed takeover of a critical production system, accomplishing this in about two days.
AVAILABILITY · Integrated into Anthropic's OSS Scanner service− Claude Mythos is among the models used to generate reports for Anthropic's OSS Scanner, which scans open-source projects that opt in.
AVAILABILITY · Preview access restricted+ Anthropic had a Mythos preview that was restricted to participating organisations.
CAPABILITIES · Reports sent without human review; include reproducers, explanations, identified changes and candidate patches+ Reports generated by Anthropic's models including Claude Mythos are sent to maintainers without human review and include a self-contained reproducer, an explanation of the suspected vulnerability, bisection to identify the introducing change when possible, and a candidate patch.
CAPABILITIES · Used to generate OSS Scanner security reports+ Anthropic's OSS Scanner generates periodic security reports for enrolled open-source project maintainers using the company's strongest models, including Claude Mythos.
LIMITATIONS · OSS Scanner reports are fully model-generated without human review and may contain errors+ Anthropic says OSS Scanner outputs — generated by models including Claude Mythos — will be fully model-generated without human review, which means reports may be incorrect or invalid.
PRICING · OSS Scanner is offered at no cost to participating open-source projectsAnthropic’s OSS Scanner offers thorough, periodic security scans by its strongest models to opted-in open-source projects at no cost.→Anthropic's OSS Scanner, which uses its strongest models including Claude Mythos, will be provided to opt-in open-source projects free of charge.
AVAILABILITY · Integrated into Anthropic's OSS Scanner service+ Claude Mythos is among the models used to generate reports for Anthropic's OSS Scanner, which scans open-source projects that opt in.
CAPABILITIES · Used as one of Anthropic's strongest models for vulnerability scanning+ Anthropic says OSS Scanner's vulnerability reports are generated by its strongest models, including Claude Mythos.
LIMITATIONS · OSS Scanner outputs are fully model-generated without human review and may be incorrect+ OSS Scanner’s outputs will be fully model-generated without human review or triage, which means it is possible the reports will be incorrect or invalid.
PRICING · OSS Scanner offered at no cost to open-source projects+ Anthropic’s OSS Scanner offers thorough, periodic security scans by its strongest models to opted-in open-source projects at no cost.
CAPABILITIES · Described as one of Anthropic’s 'strongest models'+ Claude Mythos is described in reporting as one of Anthropic’s “strongest models.”
CAPABILITIES · Used to generate OSS Scanner vulnerability reports+ Anthropic’s OSS Scanner generates vulnerability reports using its strongest models, including Claude Mythos, to scan open-source projects.
CAPABILITIES · Software vulnerability detection (claimed)+ Anthropic claimed that Claude Mythos "is better at finding software vulnerabilities than most security experts" (reported at the end of April).