Does this send my usage data anywhere?
No. Calculations run deterministically in your browser.
How estimates work
Token quantities are multiplied by the selected model's published per-million-token rates. Cached input discounts are applied only to the cached share.
Keep pricing current
Preset pricing was reviewed July 12, 2026. Switch to Custom pricing at any time to use a newer vendor rate.
Compare models side by side
Toggle between presets to compare input, cached-input, and output costs for the same prompt. The difference usually comes from output volume and cache rates, so inspect each line instead of relying on the total alone.
Plan a budget before you build
Estimate a realistic weekly prompt volume, then multiply it by the per-run cost to see monthly spend. Because the calculator is deterministic and runs locally, you can paste real prompts and log sizes without worrying about data leaving your machine.
Related tools and pages
- Compare LLM API costs
- Estimate coding-agent cost
- Use custom model pricing
- Cosmo Wise Tools - Browse more free browser-based tools.
- LLM Learning Planner - Plan a structured LLM-assisted study program.
- LLM Content Planner - Turn a content brief into a complete LLM content plan.
- Token Usage Optimizer - Find ways to reduce token usage in prompts.
- MCP Server Pricing Calculator - Estimate MCP server operating and scaling costs.
- Privacy
- Contact
Frequently Asked Questions
How is cached input calculated?
The cached share uses the model's cached-input rate; the remaining share uses its standard input rate.
Which costs does the estimate include?
Input tokens, cached-input tokens, and output tokens at the rates you configure per model. You can add your own custom model pricing when a preset does not match your deployment.
Can I use it for agent or workflow budgets?
Yes. Estimate a single agent run, then multiply by expected daily runs to project weekly or monthly spend. The deterministic local engine makes the same numbers reproducible for a planning document or a team review.