pricing
Two prices are fixed
A sample is free. A pilot on your own claim is $5,000, credited against your next engagement. Everything above that we scope with you, because two engagements at the same tier can differ by an order of magnitude.
The free sample: one repo, one claim, one signed receipt
We pick the claim. We run the verifier against the real effect, sign the verdict, and publish the receipt. You check it in your browser with only our public key.
The claim is ours. That is what makes it free. When you want your own system verified, that is the pilot below.
The pilot: your claim, one repo, one signed receipt
Credited in full against your next engagement, so it works as a deposit.
We verify one claim on your system against your own records. Thirty to sixty days, success criteria written down before we begin. With a written report and a Q&A window, $7,500.
Tell us the claim. We come back within one business day with what we would verify and what it costs. Or write to sales@turiyahq.com.
engagement --models
Three ways to buy the same engine. These are engagement models rather than size tiers, so we scope each one with you.
A single verified answer, delivered as evidence an underwriter or auditor can review.
- Independent re-check of each claim against the real effect
- Batch verification across a defined claim set
- Z3-checked formal specs where the property is provable
- Proof-carrying receipts: Ed25519-signed, content-addressed
- A written memo per claim: what was asserted, the evidence used, the attacks attempted, and what the verdict does not establish
- An assurance report across the claim set. Read the published one
The same engine on a cadence, because the mandate is a cadence.
- Re-verification every quarter, on the claim set we agreed
- Drift detection: what changed between cycles
- A signed receipt per verdict, every cycle
- Named contact, priority response
- Evidence pack mapped to AIUC-1 third-party-eval requirements
A verification programme built around your fleet, your specifications, your constraints.
- Unlimited systems and claim surfaces
- Deployment inside your environment
- Custom verification specifications
- SSO / OIDC, RBAC, audit logging
- Air-gapped deployment
- Uptime and response SLAs
pricing --drivers
What sets the price
No calculator, because there is no formula. These are the five things that actually move the number, so you can estimate before we talk.
proof
Every verdict is Ed25519-signed and content-addressed. Verify any of them in your own browser. No trust required to verify the record. And read how we verify, including the candidates we rejected: that is the part a vendor inflating a record cannot show you.
faq
Why are only two prices on this page?
Because those are the two with a fixed scope. A sample is one claim on a repo we choose. A pilot is one claim on yours, bounded at thirty to sixty days. Everything above that varies by how many systems are in scope and how much of the specification has to be built, so we scope it with you instead of quoting a number we would have to take back.
What does the free sample include?
One repository, one claim, verified end to end, and a signed receipt you can check in your browser with only our public key. No credit card, no call required. The claim is ours. We pick a system we can verify against an independent record and publish the result, so you can watch the method work before you spend anything. A signature is not evidence of correctness, so read the receipt and decide for yourself.
What does the $5,000 pilot buy?
One claim on your system, verified against your own records, with success criteria agreed in writing before we start. Thirty to sixty days. The fee is credited in full against your next engagement, so it works as a deposit. If the finding is not what you hoped, you keep the receipt and the finding either way. That is the point of running it.
What is the quarterly programme for?
AIUC-1 certification requires third-party technical re-testing every quarter: a permanent obligation for as long as a system stays certified. The quarterly programme is built around that cadence: re-verification each cycle, a signed receipt each time, and an evidence pack mapped to the standard.
Do you white-label the engine?
Yes: for MGAs, accredited auditors and red-team firms who need verification capacity they do not staff. That is a different arrangement with its own page: see Partners.
Can we verify our own systems, or do you need access?
You can. The engine runs against your repo, your test, or your specification. Your model and data never leave your environment unless you ask us to run a hosted verification on your behalf.
How do we start?
Start by reading a sample. We pick the claim, verify it, and publish the receipt, so there is nothing to sign up for. Or send one message describing what you want verified, and we will scope a pilot on your own system.
Start with the receipt, or start with a conversation.
Either works.