On September 17, 2026, Anthropic introduced the Life Sciences Verification Program (LSVP), a beta initiative that grants verified life science organizations access to its Mythos, Opus, and Sonnet model families with safeguards tuned for biology workloads. The program opens applications to the broader community after onboarding dozens of organizations in early access.

Related: Earlier coverage: Anthropic ships Claude Fable 5.1 and Mythos 5.1 with 75% cheaper cache reads.

LSVP offers two grant tiers. Standard Use grants cover most biology R&D workflows — basic science, drug discovery, clinical development, manufacturing, quality assurance, regulatory affairs, and investing — and are renewable annually for entire teams. High-risk Use is an add-on for work blocked under Standard Use; it removes all life-science content blocks, applies to a single research project, and renews every six months. High-risk grants for Opus 5 and Sonnet 5 are available at launch, while Mythos high-risk access remains limited to a small set of entities pending coordination with the U.S. government.

Confirmed

  • Program name: Life Sciences Verification Program (LSVP), launching in beta on September 17, 2026.
  • Models included: Mythos 5.1, Opus 5, Sonnet 5, and future models as they launch.
  • Grant types: Standard Use (team-wide, annual renewal) and High-risk Use (project-scoped, six-month renewal).
  • Verification requirements: review of research credentials, security standards, and ethical research oversight.
  • Access surfaces: API console, Claude for Enterprise, Claude for Teams, Claude Science, Claude.ai, and Claude Code.
  • Current limitations: individual Pro/Max plans not yet supported; third-party platforms not supported; BAA-enabled orgs excluded (PHI data requires separate non-BAA orgs).
  • Data retention: 30 days for LSVP traffic to enable offline monitoring; data compartmentalized, not used for training, not accessible to Anthropic life sciences researchers.
  • Early-access partners named: Xaira Therapeutics, Edison Scientific, Manifold Bio.
  • Enrollment target: hundreds of organizations in the first week, scaling to majority of life science community in coming weeks.

Unknown

  • Exact criteria and documentation required for verification (beyond the three categories stated).
  • Pricing or cost structure for LSVP grants versus standard API/Enterprise plans.
  • Independent audit or third-party oversight of the offline monitoring system.
  • Timeline for individual Pro/Max plan access, third-party platform support, or BAA enablement.
  • Specific U.S. government agencies involved in Mythos high-risk gating and the vetting process for the "small set of entities."
  • Whether the 30-day retention window is fixed or adjustable per organization.

Our take

Anthropic is betting that institutional vetting plus behavioral monitoring can replace blunt content filters for biology workloads — a shift from real-time blocking to post-hoc pattern detection that mirrors enterprise security operations. The model makes sense for organizations with mature oversight, but the six-month High-risk renewal and government-gated Mythos access signal that the dual-use boundary remains contested. The real test will be whether offline monitoring catches diverted access faster than the previous filters blocked it, without chilling legitimate research.

Sources