Estimate the input tokens and cost of sending an image to an OpenAI vision model. Select a model, enter your image dimensions, and choose a detail level.
For GPT Image generation and editing costs, use the image generation calculator.
Per image, at standard input rates. API pricing
Calculation details
Fits within 65,535 × 65,535 pixels. Images are not resized to fit the patch limit. Smaller images are not enlarged.
The API rejects images above 30,000 patches after resizing. See the image input limits.
- Size after resizing
- 1024 × 1024 px
- 32 × 32 patches
- 32 × 32 = 1,024
- Token multiplier
- 1.2×
- Input price per 1M tokens
- $4.00
Excludes other prompt tokens, caching, long-context pricing, and data-residency adjustments. Billing can differ by one token due to rounding.
Use the calculator
- Select the vision model you plan to use.
- Enter the original image width and height in pixels. The calculator applies the model’s resizing rules.
- Select an image detail level supported by the model.
- Read the image input tokens and estimated cost. If the processed image exceeds the 30,000-patch limit, the calculator shows a rejection message instead of an estimate. Expand Calculation details to see the resized dimensions and token calculation.
For example, a 6000 × 6000 image on GPT-5.6 exceeds the limit with original detail (35,344 patches), but fits after resizing with high detail (2,500 patches). Choose high only when your task does not require original resolution or precise image coordinates.
Understand the estimate
The estimate covers one image at standard input rates. It excludes other prompt tokens, model output, caching, long-context pricing, and data-residency adjustments. Billing can differ by one token due to rounding.
For the resizing and tokenization rules, see image input cost calculations. For current model rates and other charges, see API pricing.