### Answer Simulation tests bill per message, not per test, so the cost of 170 tests depends on how many messages each run produces and which models are involved. How to calculate it - A batch bills every case on every run, with no discount. 170 cases run once is 170 runs, and running them three times is 510 runs. [(1) ]() - Each simulated run bills two models: the model playing the simulated user (the LLM you pick in the test case) and the model your agent uses, set in the agent's own LLM settings. [(2) ]() [(1) ]() - Message rates range from roughly $0.001 to $0.05 per message depending on the model. [(1) ]() - Each case in a batch also adds one post-call-analysis unit for grading the pass or fail verdict. Formula Total ≈ (number of runs) × (messages per run) × (per-message rate of the agent's model + per-message rate of the simulated user's model) + (number of runs × grading unit cost) Practical steps 1. Run one or a few of your test cases and check

Billing → Usage

to see the actual cost per run. 2. Multiply that average by 170, and by the number of times you plan to rerun the batch. 3. Check the [Retell pricing page]() for the current per-model rates and the grading unit cost. A cheaper model for the simulated user keeps costs down, but it doesn't help if your agent runs an expensive model, since the agent's replies bill at its own model's rate.

​ To learn more, see the sources I considered: [[1] Testing pricing](https://docs.retellai.com/test/testing-pricing) [[2] Automatically test your agent](https://docs.retellai.com/test/llm-simulation-testing) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.