# How is inference cost per token calculated?

Inference cost per token is the provider’s price per million tokens, split by direction: input tokens (your prompt) and output tokens (the response), with output typically 2 to 6 times dearer. A request’s cost is input tokens times the input rate plus output tokens times the output rate.

A worked example at catalog list prices: a request with 1,200 input and 350 output tokens on Gemini 3.5 Flash ($1.50 in, $9 out per million) costs 0.0012 × $1.50 + 0.00035 × $9, about $0.0049. A million such requests is about $4,950 a month.

Three modifiers move the real number: cached input bills at a published discount, batch traffic can run on cheaper models, and routing sends each request shape to the model priced for it. On Allocate the billed price is always list plus the 7% transaction fee.

## See also

- [LLM price comparison](https://allocate.network/tools/llm-price-comparison)

## Related terms

- [Model routing](https://allocate.network/glossary/model-routing.md)
- [Tokens](https://allocate.network/glossary/tokens.md)
- [Prompt caching](https://allocate.network/glossary/prompt-caching.md)

---

[HTML page](https://allocate.network/glossary/inference-cost-per-token) · [Machine-readable catalog](https://allocate.network/catalog.json)