Every feature on both plans
The plans differ in how much you use, not in what you can do.
| Prices in US dollars. Cancel any time. | Free$0 a monthStart for free | Pro$39 a monthContinue with Pro |
|---|---|---|
| Usage | ||
| Projects | 1 | 5 |
| Tools | Unlimited | Unlimited |
| Evals in a project | 20 | Unlimited |
| Eval runs a month | 200 | 2,000 |
| Tool calls a month | 10,000 | 100,000 |
| Call history kept | 7 days | 30 days |
| Product | ||
| Import an OpenAPI document | Included | Included |
| Connect an existing MCP server | Included | Included |
| Hosted MCP for your tools | Included | Included |
| Edit what agents read | Included | Included |
| Evals on Production and the draft | Included | Included |
| Versioned publishing | Included | Included |
| Tool call log and regression tests | Included | Included |
| Support | ||
| Email support | No | Included |
Good to know
- What counts as an eval run
- One eval checked against one version. Running 10 evals against your draft is 10 runs. Comparing the same 10 against Draft and Production is 20.
- The same model on both plans
- Free and Pro evals use the same model, so a result means the same thing on either plan. You do not bring your own API key.
- When a run counts
- A run counts when Riviera asks the model, even if the model fails to answer. A run that never reaches the model isn’t counted, including when model tests are temporarily unavailable.
- When you reach a limit
- Usage is blocked until the next month or until you upgrade. There are no overage charges. If Pro is not enough, email hello@riviera.so.
- Your data on every plan
- Call inputs and results are stored only if you turn capture on, and it is off by default. The credential for your server is encrypted and never shown again. Your server still runs every call.