AI API price calculator

Pick a model, enter your expected monthly volume and see what it costs with us — next to the same volume at the model maker's official price.

Model

GPT-6 Solgpt-6-sol▾

ChannelDefault (Private Pool 3) — $0.1096 / $0.548codex — $0.048 / $0.24Private Pool 5 — $0.30 / $1.1334

Channels are different supply routes for the same model. Without a suffix your requests use the default one.

Input tokens / month

M

Output tokens / month

M

Cache hits (share of input)50%

Cached input is billed at ×0.1 of the input price.

Count cache writesUncached input is billed as a cache write at ×1.25 of the input price.

GPT-6 Sol

Default

Your cost per month

$1.70

At the official price

$30.81

You save

$29.11(−94%)

Line	Per 1M	Cost
Input, uncached5 M	$0.1096$2.02	$0.5480$10.10
Input, cache read5 M	$0.01096$0.101	$0.0548$0.5050
Output2 M	$0.548$10.10	$1.10$20.20

Upper line — our price; struck-through — the official price for the same tokens.

How it is calculated

• Input cost = uncached input tokens ÷ 1M × the input price per 1M, plus cache-read tokens ÷ 1M × the cache-read price. Output cost = output tokens ÷ 1M × the output price.
• The cache share splits your input tokens: that part is billed at the channel's cache-read price (its multiplier of the input price), the rest at the full input price.
• With cache writes on, the uncached input is billed at the channel's cache-write multiplier — this is how Claude prompt caching is charged.
• The official price is the model maker's published list price per 1M tokens in USD, plus 1% for payment fees — the same reference as the Savings column on the models page. Where the maker lists no cache-read price, cached input is priced with the channel's multiplier, so the saving is never overstated.
• Saving = official cost − our cost; the percentage is taken of the official cost. Prices are the live catalog prices on this page.

Get an API keyAll models and prices