Private AI cost tool
AI Token Cost Calculator
Estimate per-request and total API costs from uncached input, cached input, output tokens, and request volume. Calculations stay in your browser.
Provider prices verified 2026-07-29. Check the linked official pricing source before making purchasing decisions.
How token cost is calculated
Each token category is multiplied by its price per one million tokens. The calculator adds uncached input, cached input, and output costs, then multiplies the per-request total by the request count.
Built-in provider prices are a dated snapshot, not a live billing quote. Qwen presets use CNY and apply documented input-length tiers. Use Custom rates for promotions, regional pricing, negotiated rates, or newly released models.
Does this upload my files?
No. Everything on this page runs locally in your browser, so your files are never uploaded to a conversion server. Close the tab and nothing is left behind.
After any required processing engine has loaded, repeated conversions can keep working offline. You can verify the claim with the site's reproducible browser privacy audit.
FAQ
What does cached input mean?
Cached input tokens are reused prompt-prefix tokens billed at a discounted model rate. Enter them separately from ordinary uncached input tokens.
Does this connect to my API account?
No. It performs multiplication locally using the token counts and request volume you enter. It cannot read your account or billing data.
Can I calculate another model or provider?
Yes. Select Custom rates and enter the provider's current input, cached-input, and output prices per one million tokens, then choose the matching currency.