How the cost is calculated
AI providers bill input and output tokens separately, usually per million tokens (MTok). Input tokens are your prompt, system instructions, and any context; output tokens are what the model generates in response.
Why output costs more than input
Every provider prices output tokens several times higher than input — typically 5x to 6x. Processing input can happen in parallel in a single forward pass through the model, while generating output requires a separate pass per token, which is far more compute-intensive.
A note on the presets
Pricing shown is current as of July 2026, sourced from each provider's public pricing page. AI pricing changes often — sometimes monthly. Always double-check against the official pages before budgeting a production workload: Anthropic, OpenAI, Google Gemini.
Common uses
- Budgeting a feature: estimating the monthly cost of an AI feature before shipping it.
- Comparing models: checking whether a cheaper model meaningfully changes your cost at your expected volume.
- Sizing a request: understanding how much a single long-context call costs before sending it.
Frequently asked questions
Does this include prompt caching or batch discounts?
No — this calculates standard, non-discounted pricing. Prompt caching (often 90% off cached input) and batch processing (typically 50% off) can significantly lower real costs; adjust the input/output prices manually to model those scenarios.
How do I estimate tokens from a piece of text?
As a rough rule of thumb, one token is about 4 characters of English text, so 1,000 tokens is roughly 750 words. For exact counts, use the tokenizer tool published by each provider.