Run established benchmarks or bring your own benchmark (BYOB).
Run evaluations in parallel without managing the infrastructure.
Set a spending cap before each evaluation. Track what you spend as it runs.
Compare model scores and costs. Inspect answers and export reports.
Pricing.
No subscription. Buy credits and pay per run.
Pay as you go
12.5%platform fee
Applied to compute and inference, with a $1 minimum per model evaluation. Usage costs are additional.
- Established benchmarks and private Hugging Face imports
- Concurrent model calls and sandboxed code grading
- Managed models at provider cost, or your own endpoint
- A spending cap on every evaluation
- Model comparisons, answers, and JSON, CSV and HTML exports
- The CLI, HTTP API and agent setup
Enterprise
Custom
Everything in Pay as you go, plus
- Custom pricing
- Direct support from the team that runs EvalRouter
How is the fee calculated?
12.5% of compute and inference, with a $1 minimum per model evaluation. Usage costs are additional.
What does the spending cap cover?
The spending cap covers EvalRouter charges, including its platform fee. Connected model providers bill inference separately.
