Pricing and process
Starting prices for every way to work with Vencur. The final quote depends on the number of tasks, the format and the workflow.
-
Free sample
Free
One task, runs locally in minutes.
-
Most teams start here
Evaluation pack
From $6,000
25 tasks in your format to test your model, invoiced only after you accept it.
-
Training licence
From $600 per task
Non-exclusive by default; typical packs of 25–80 tasks. Exclusive licences on request.
-
Custom world
From $40,000
A new world for your workflow, with its own verifier and task generator.
-
Quarterly refresh
From $20,000 per quarter
New, harder tasks every quarter, so your environment never saturates.
-
Pilot
From $24,000
A scoped pilot for one workflow before a larger order.
Ready-made packs
Ready-made task packs from OpsDesk-Care 2.4.13. Every task has a CP-SAT reference and passes our QA and red-team checks.
-
Starter 10
10 tasks across all task families. The fastest way to try the world.
$2,700
-
Honesty 50
50 tasks where full coverage is provably impossible or only partly possible. Measures whether an agent reports honestly.
$10,000
-
Hard Eval 100
100 hard and very hard tasks for comparing frontier models.
$25,000
-
Curriculum 200
200 tasks from easy to very hard for training. Generated on order in about one week.
$35,000
-
Volume
1,000+ tasks, your mix of families and difficulty, with an optional quarterly refresh.
From $100,000
All packs are non-exclusive. Exclusive licences on request. Invoiced after acceptance.
Delivery, licence and invoicing
-
Delivery
An evaluation pack is typically delivered in about ten working days, as files or as a hosted environment.
-
Licence
Licences are non-exclusive by default. Exclusive licences are available on request.
-
Invoicing
You are invoiced after you accept the delivery, not before.
At a glance
| Offer | Price | Scope | Best for |
|---|---|---|---|
| Free sample | Free | One task | A first look at the format |
| Evaluation pack | From $6,000 | 25 tasks in your format | Testing your model |
| Training licence | From $600 per task | Typical packs of 25–80 tasks | Training on verified tasks |
| Custom world | From $40,000 | A new world with its own verifier | A workflow not covered yet |
| Quarterly refresh | From $20,000 per quarter | New, harder tasks every quarter | Keeping a world from saturating |
| Pilot | From $24,000 | One workflow, scoped together | Trying a workflow before a larger order |
| Starter 10 | $2,700 | 10 tasks across all task families | Trying the world |
| Honesty 50 | $10,000 | 50 tasks with impossible or partial full coverage | Measuring honest reporting |
| Hard Eval 100 | $25,000 | 100 hard and very hard tasks | Comparing frontier models |
| Curriculum 200 | $35,000 | 200 tasks from easy to very hard | Training agents |
| Volume | From $100,000 | 1,000+ tasks, selected families and difficulty | Large-scale training and evaluation |
Questions about pricing
Why do the prices say “from”?
Each price is the starting point for the smallest scope. The quote depends on the number of tasks, the format and, for custom worlds, the workflow.
When do we pay?
After you accept the delivery. You are invoiced only once you have checked it.
Is the licence exclusive?
Non-exclusive by default. Exclusive licences are available on request.
Can we start small?
Yes. The sample task is free, and an evaluation pack lets you test your model on 25 tasks before you license more.
Which supplier details do you provide?
The procurement page lists the supplier facts: EU-based, synthetic data only, delivery without access to your systems. A US tax form W-8BEN is available on request. Procurement