AI COST CALCULATOR
Same workflow.
Very different bill.
Estimate and compare AI model costs instantly. Enter a prompt or build a workflow, then see how costs differ across Claude, GPT, DeepSeek, and more.
Example: support reply
Examples:
Estimated from text. Edit for accuracy.
Includes any reasoning tokens billed.
Build a workflow with 1-5 steps. Each step is a separate model call. Each comparison row prices every step using that model.
Estimated Costs
Loading pricing data...
Unable to load pricing data.
| Model | Per run | Per 1,000 | Monthly | vs baseline |
|---|
Estimates only. Same token assumptions, different model prices. Quality varies: run a blind test.
Monthly estimates apply the current UTC quote to every run; scheduled prices may change.
Assumptions and limitations
- Costs are estimates only based on stated per-token pricing.
- Token counts are approximations (~4 chars/token for English). Actual tokenization varies by model.
- Reasoning tokens are billed as output tokens where applicable.
- Caching discounts, batch processing, web search, image/audio, and provider fees are not included.
- DeepSeek uses time-based pricing (prices shown for current UTC time; monthly projections assume constant rates).
- GPT Astra and Luna have tiered pricing above 272,000 input tokens per step.
- Cost comparison does not imply equivalent quality. Run a blind test to compare outputs.
- Pricing source: OpenRouter pricing overrides.
Step breakdown
Measure your real costs
These are estimates. With Metergraph, track actual spend across models and routes in production.
Sign up for MetergraphOr run a blind quality test to compare outputs.