Pay per audit credit
Audit is the billing unit: not seats, not API calls, not storage. One credit is one audit report. You pay for audit reports produced, nothing else.
Sign up with no invite code and no human contact. Get an API key and your first audit the same day.
- 25 audits / month
- Open agent_owner scope, no pre-registration
- Full audit report with PDF export
- Hash chain: tamper-evident step hashes
- LangChain and CrewAI adapters, plus a CI gate action
300 audit credits per month (1 credit = 1 audit) for teams building their first AI accountability programme.
- 300 audits / month
- Everything in Free
- Standard rate limit (60 req/min, up from 10)
- Priority quota reporting
2,500 audit credits per month for teams running Jiminy in production.
- 2,500 audits / month
- Everything in Starter
- Agent registry, portfolio/estate-level rollup
- Closed-loop remediation workflow
- Benchmarking & normalcy context on verdicts
Review before rollout
Fixed fee, agreed before we start.
For a team about to put an agent into production. An independent Agent Audit of one agent on test or shadow data, a short written report, and a walk-through with your team and, if you want, your assurance team.
Pricing reflects independence
Jiminy only charges independent evaluators, not the organisations whose agents are being evaluated. Agent owners submit traces; evaluators run audits and pay for the reports. This keeps the billing relationship aligned with accountability: the party producing the audit is the party paying for it.
On independent evaluator accounts, which are set up on request, no: the submitter and agent owner are enforced as different organisations at the account level, not by policy or payment configuration. The Free and Starter self-serve tiers work differently by design: they are for teams auditing their own agents before they need independent sign-off, so one account can hold both roles and results are self-declared rather than organisationally verified. If you need independence for a compliance or audit context, talk to us.
One audit report corresponds to one DecisionTrace evaluated to a final verdict. Multi-run audits (up to five independent passes) count as a single report; the runs are how we produce a reliable verdict, not additional billable units.
Yes, the Free self-serve plan includes 25 real, persisted audits per month with no invite code and no charge. Separately, calibration mode (available on request) lets you run the judge against your traces without persisting results or receiving a formal report, useful for onboarding before your first billed audit.
Standard retention is 6 months from the audit date. Audit records are retained for 10 years, automatically and at no extra cost, where extended retention applies and for domain profiles we classify as high-risk (currently: healthcare prior-authorisation, financial trading, and insurance underwriting; see Security for the full retention table). If your use case doesn't fall into one of those categories but you expect a liability, regulatory, or audit claim could plausibly surface more than 6 months after a decision, ask us about extended retention before you rely on the standard default: a claim under an evolving liability framework like the EU Product Liability Directive can in some circumstances be brought years after the decision that gave rise to it, well outside a 6-month window. We're not your lawyers and this isn't legal advice about your specific exposure. It is a plain statement that the 6-month default is a cost/scope tradeoff, not a claim that 6 months is always enough. Extended retention exists precisely for the cases where it isn't.
Independence where it matters.
Self-serve sign-up is open now, no invite code required. Going live with an agent soon? Talk to us about a review before rollout.