Online… visits today
Workload assessment
Is the model you chose still the right one?
A fixed-scope assessment of one workload against three candidate models, using your tasks rather than ours. You get a decision report within seven business days of us having what we need — including, where the evidence supports it, a recommendation to change nothing.
$1,290one-off, fixed scope · USD, excluding any applicable tax
- 20
- of your own tasks at most, with the answer keys held out of the prompts
- 3
- candidate models, several trials per task
- 7
- business days from us having your tasks to the decision report
Scope
What is included
- One workload, up to 20 tasks you supply, with the answer keys held out of the prompts
- Three candidate models, up to 1,000 standard execution units, several trials per task
- Two hours of expert time, and a concise decision report
Process
How it works
- Book and pay below. Payment is by card through Stripe, and an invoice with your company details is issued automatically.
- We confirm the scope within two business days — the tasks, the three models and what “good” means for your workload.
- You send your tasks. We build the suite and run it repeatedly against the candidates.
- You get the report within seven business days of us having your tasks.
Guarantees
If it does not work out
- If, once we have read your workload, we cannot measure it, we refund the full amount before any work starts.
- If we cannot deliver the agreed report, we refund in full.
- The full amount is credited against an annual plan bought within 30 days, up to that plan’s price — with Teams, the year costs nothing extra.
Provider inference is billed to your own API keys under a cap you agree first, so the cost of running the models is never hidden in our price. Your tasks stay yours: never published, never added to the public benchmark.