Use the Estimate AI Cost action before calling an AI provider.
Choose a provider and model, then enter the prompt text, expected response tokens, max budget, and safety buffer. The action returns estimated input tokens, estimated output tokens, input cost, output cost, total cost, cost with buffer, whether the request is within budget, risk level, and a recommendation.

Example workflow:
Step 1: Estimate AI Cost
Step 2: Only when Result of step 1's within_budget is yes → Call your AI provider
Step 3: Only when Result of step 1's within_budget is no → Show an alert or stop the workflow

Recommended expected response token values:
300 = short response
800 = medium response
1500 = long response
3000+ = detailed response or report

Use custom input/output price fields only if you want to override the built-in model pricing. Leave them blank for normal use.

Note: estimates are approximate because each provider uses its own tokenizer and pricing can change over time.