Watch: cut an image's token cost
No sound. Each step is captioned on screen and also written out in the guide below.
AI Tools
See how many tokens an image costs on GPT, Claude, and Gemini, using each provider's own rules. Then resize it, and see what you'd save before you send it.
Add an image to see its token cost.
| Model | At original size | At new size | Saved |
|---|
Rules from each provider's documentation. Gemini 3 and later charge a fixed amount per image, set by media_resolution: 280 (low), 560 (medium), 1,120 (default), or 2,240 (ultra high).
No sound. Each step is captioned on screen and also written out in the guide below.
Drop, choose, or paste a screenshot, photo, or scan.
See its token cost on each one, and the size each model actually looks at.
Pick a smaller size, check the text is still readable, and send that instead.
Sending a bigger image than a model looks at only makes the upload larger; sending a smaller one can cut tokens by half or more.
They follow each provider's published rules (checked October 2026) and match their worked examples. Providers change these rules between models, so for billing, trust the usage numbers the API returns.
It depends on the image. Photos stay clear at much smaller sizes; screenshots and scans need their text to stay readable. Check the preview: if you can't read it, the model may not either.
Gemini 3 and later charge a fixed amount per image, set by the media_resolution option. To pay less there, choose a lower setting, such as low (280 tokens) for simple images.
No. Measuring and resizing both happen in your browser. The resized image is drawn fresh, so location and camera details aren't copied into it.