Pricing
Fixed, published, and never success-based.
Every fee below is the fee. Nothing is quoted, nothing is negotiated, and no price ever depends on what the measurement finds. That discipline is part of what makes the evidence worth having.
FIVE BUSINESS DAYS · NO MEETINGS, EVERYTHING IN WRITING · NO CUSTOMER DATA · WRITTEN AGREEMENT FIRST
For organizations that run AI
Measurement of the model you host or control.
LLM Validation Report
$9,500, one model configuration, signed and dated, five business days from materials-complete. Founding program: $4,500 for the first three clients, once per client, in exchange for named case-study permission, with quarterly re-measurement locked at founding rates for two years.
Continuous Assurance
$6,500 per quarter for one production model, $8,000 per quarter for a second model or a heavy change cadence. The quarterly trend record, pre-deployment change gates, and an incident allowance live here and are not purchasable piecemeal.
Model Selection Study
$4,500 for two candidate models, plus $1,500 per additional candidate, to a maximum of four candidates for $7,500. The cheapest moment to measure is before you commit.
For companies that sell AI
Independent evidence your buyers will actually accept.
The Vendor Assurance line serves AI product companies whose enterprise buyers, hospital committees, and insurers now ask for reliability evidence the vendor did not grade itself. Details on the AI vendors page.
Vendor Evidence Pack
$12,500, one product and one model configuration: the signed evidence report, a shareable summary letter, a questionnaire-ready evidence brief, and twelve months of written answers to your buyers' reviewers. Founding program: $6,500 for the first three vendors, invoiced on delivery, in exchange for named case-study permission.
Scope, not negotiation
Additional model or deployment configuration +$4,000 · expedited 48-hour delivery +$3,000 · post-remediation re-measurement within 60 days $3,500 (included once at the founding rate).
Evidence Refresh
$5,000 per quarter keeps the evidence inside the freshness window enterprise reviews ask about; founding vendors may lock $4,000 per quarter for two years.
Training-Data Ingestion Screen
$8,500 per training event standalone, $4,500 alongside an Evidence Pack: a ratio-based screening measurement of whether a fine-tune absorbed a specified corpus, with positive and negative controls and its limits stated plainly.
CHAI Applied Model Card supplement
$2,500: measured results formatted into the Applied Model Card fields health-system buyers request.
Straight answers
Questions about the numbers
Why publish prices when nobody else does?
Because the buyers this lab serves are compliance and procurement professionals, and unpublished pricing costs them a discovery call they do not want. Fixed, published fees are also part of the independence story: the fee never depends on what the measurement finds, and it never moves in a negotiation.
Do prices ever get discounted?
No. The founding programs are the only reduced rates this company will ever run, they exist in fixed numbers of slots, and each carries its stated reason. Larger engagements cost more through scope, additional models, configurations, or expedited delivery, never through negotiation.
What is the founding trade, exactly?
The founding rate is exchanged for permission to describe the engagement in a short named case study, subject to the client's written approval of the exact text. It is not a testimonial trade. The rate is low on purpose: the quarterly re-measurement relationship is the real business, and the first clients make it possible.
What are the payment terms?
Written engagement letter before any work begins, invoice on delivery terms stated in the letter, ACH or check. Fees are never contingent on results, and repeat one-off measurements bill at list price unless covered by a quarterly program.
The sample report shows what the money buys.
Read the full sample first, then write to the lab with your question types and endpoint. Everything is answered in writing, on the record.
Read the sample reportOR EMAIL CONTACT@SPECTRALGRAPH.AI