Start free. Pay for what you use.
Sandbox usage.
| Resource | Hourly rate |
|---|---|
| Default CPU | $0.06 / vCPU-hour |
| Default memory | $0.018 / GiB-hour |
| Pinned CPU | $0.15 / vCPU-hour |
| Pinned memory | $0.05 / GiB-hour |
Usage credit covers sandbox usage and eligible runtime overage. Runtime-only credit is used for runtimes first. Credits do not roll over; activation and plan changes can prorate grants.
A monthly allowance, plus team features.
Local scanning, plus card-verified hosted credits.
Monthly hosted credits for security work.
Monthly hosted credits for security work.
Custom commercial terms for security teams.
Bounded runtimes, signed evidence.
Explicit target, region, policy, and budget on every runtime, with deny-default egress and approval gates for credentialed, destructive, and exploit-class actions.
Coverage across prompt injection, tool abuse, MCP trust-boundary failures, permission escalation, and data disclosure, mapped to OWASP LLM Top 10 and MITRE ATLAS.
Every runtime emits a signed evidence bundle you can verify with tachyonic verify.
Paid plans run in US and EU regions; team plans add shared findings, reports, reviewer share links, and approval workflows.
Start your first runtime.
Local scanning requires no card. Hosted Free credits require a verified card.