Pricing Made Simple.
Tell us what your agent does and share some example data. We write the tests, find where it could go wrong, and give you a clear report on what to fix.
Free Tier
Everything you need to test an agent, with the evals built for you. A great way to try LuminosAI on a real use case.
Each one built for your agent and run on your data
Your 10 evals quota resets each month
See what passed, what failed, and what to fix
Pay-as-You-Go
Buy credits up front and use them whenever you need to test. Good when testing comes in bursts, like before a launch.
Run a new eval whenever you change your agent or launch a new one
Run evals as often as you need, on test data of any size
Run more evals at once when you connect through our API
Top up credits when you need more. No contract required.
Subscription
A set number of evals every month at a lower price per eval. Built for teams that test regularly, on new agents or ones already in use.
A set number of evals included every month
Pay less for each eval than with prepaid credits
Choose how long the data you upload is stored
Our highest limits, for testing every time your team ships a change
Enterprise
For organizations testing many agents or with special requirements. Includes expert help and dedicated support.
Pricing built around how many agents and teams you test
Lower rates and custom terms for testing many agents
Our team adds human review and guidance to your automated evals
SSO / SAML with your company login at no extra cost
You build the agent.
We build the eval.
Most eval tools are built for engineers. They expect you to write code, build your own tests, and guess how your AI might fail. With LuminosAI, you don't need to code. We design the tests, check your data, and run everything for you.
Requires coding, engineering skills, and separate AI bills
- Requires Coding & Engineering Skills: Engineers have to write code to set up the tests, connect them to your AI, and keep it all running.
- You Have to Write the Tests: Someone on your team has to come up with hundreds of tricky test questions and decide how to grade the answers.
- Bring Your Own AI Accounts: You have to set up accounts with AI providers and keep an eye on their usage limits and bills.
- Missing What Could Go Wrong: Teams check that the agent works when things go right, but miss the unusual situations and risks where agents actually fail.
- Two Mistakes at Once: Simple tests miss real risks and raise false alarms that hold up launches.
We build the eval. No extra costs.
- No Coding Required: Describe your agent in plain English and share your data. We create the tests and find the problems you didn't know to look for.
- No AI Accounts Needed: You don't need your own accounts with AI providers or to keep track of their limits.
- No Extra Costs: Creating the tests, checking your data, and grading the results are all included. No separate bills.
Pricing that fits how your team actually works.
Buy credits as you go, subscribe monthly and add more when you need it, or sign an annual contract. We'll set it up around how often you test.
Prepaid Credits
No ongoing commitment. Buy credits up front and use them as you release new versions of your agent.
Monthly Subscriptions + Top-Ups
A predictable number of evals each month for teams that test every change. Add more anytime you need extra.
Enterprise & Annual Contracts
Annual commitments, custom contracts (MSAs), security reviews, invoicing with 30-day payment terms, and dedicated support agreements (SLAs).
Frequently asked questions.
What is included in the Free Tier each month?
10 evals every month. Each eval is built for your agent from a plain-English description and run on your data. Your evals reset each month. No credit card needed and no extra costs.
What counts as an eval?
An eval is one complete test of your agent. You describe what your agent does and share your data. We create the tests and the grading, run them, and give you a report showing what passed, what failed, and what to fix.
What does "no extra costs" mean?
Running an eval uses AI behind the scenes to grade your agent's answers. Most eval tools expect you to design the tests yourself, charge for an eval, plus require your credentials to a third party inference service. With LuminosAI, we build your evals and everything they use is included. No separate bills.
How do Pay-as-You-Go and Subscription plans handle billing?
Both are prepaid. With Pay-as-You-Go, you buy credits up front and use them whenever you need to, with no ongoing commitment. Subscriptions give you a set number of evals each month at a lower price per eval, plus sign-in with your company account (SSO / SAML). You can add more evals anytime at the discounted rate.
Can we sign an annual contract?
Yes. Annual contracts are available on our Enterprise plan. They can include volume pricing, custom contracts (MSAs), security reviews, and invoicing with payment terms. Enterprise customers can also purchase the ZwillGen Automated Legal Risk Package separately.
How do I get started with Pay-as-You-Go, Subscription, or Enterprise?
Contact us through our form and we'll get back to you quickly to set up your account.
Ready to verify your AI agents? Start testing in minutes.
Join teams testing their AI agents with no setup, no credit card, and no extra costs.