What it costs
Per million tokens (How model prices are quoted: dollars per million tokens, charged separately for what you send and what comes back.), in USD, as published..
| Charge | Per million tokens | What it is |
|---|---|---|
| Input (The unit a model reads and writes. Roughly three quarters of an English word, so 1,000 tokens is about 750 words.) | $0.1 | Every token you send. |
| Output (The unit a model reads and writes. Roughly three quarters of an English word, so 1,000 tokens is about 750 words.) | $0.5 | Every token it generates. |
| Cache read (Paying a reduced rate for a prefix the vendor has already processed, instead of full price for sending it again.) | $0.01 | An input token served from the cache instead of being charged in full. |
| Cache write, 5 minutes (The charge for putting a prefix into the prompt cache. Usually more than the input rate, not less.) | $0.125 | Putting tokens into a cache that lives five minutes. |
| Cache write, 1 hour (The charge for putting a prefix into the prompt cache. Usually more than the input rate, not less.) | $0.2 | The same, for an hour. Usually the biggest line on an agentic bill. |
What it can do
- Tool use
- Reads images
Stated as not supported: Makes images.
Not stated either way: Reads audio, Reads video, Computer use.
In practice
This section is written by us, not read off a vendor page. Everything above is not.
For high-volume, latency-sensitive tasks such as classification, extraction, and routing
These are the prices for prompts up to 100,000 tokens. Over that the page lists $0.50 input, $0.625 five-minute cache write, $1 one-hour cache write, $0.05 cache hits and $2.50 output.
Catalogue last refreshed 2026-10-10.How we verify a price.