Price list
Image models, priced per picture
Every model that draws, at the price its vendor publishes: per image, per megapixel or per token. One key calls any of them.
Each price includes Antimatter's 10% margin on the upstream's published price. A picture is billed what the upstream reports for it, plus that margin.
7 image models, cheapest picture first
Nano Banana (Gemini 2.5 Flash Image) · Google
google/gemini-2.5-flash-imageEdits from up to 3 reference images
Try in chat$33.00 per 1M image tokens
upstream $30.00 per 1M image tokens
Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) · Google
google/gemini-3.1-flash-lite-imageEdits from up to 14 reference images
Try in chat$33.00 per 1M image tokens
upstream $30.00 per 1M image tokens
Nano Banana 2.1 · Google
google/gemini-nano-banana-2.1Edits from up to 14 reference images
Try in chat$33.00 per 1M image tokens
upstream $30.00 per 1M image tokens
Nano Banana Pro (Gemini 3 Pro Image Preview) · Google
google/gemini-3-pro-image-previewEdits from up to 14 reference images
Try in chatfrom $66.00 per 1M image tokens
upstream from $60.00 per 1M image tokens
Nano Banana 2 (Gemini 3.1 Flash Image) · Google
google/gemini-3.1-flash-imageEdits from up to 14 reference images
Try in chatfrom $66.00 per 1M image tokens
upstream from $60.00 per 1M image tokens
Nano Banana 2 (Gemini 3.1 Flash Image Preview) · Google
google/gemini-3.1-flash-image-previewEdits from up to 14 reference images
Try in chat$66.00 per 1M image tokens
upstream $60.00 per 1M image tokens
Nano Banana Pro (Gemini 3 Pro Image) · Google
google/gemini-3-pro-imageEdits from up to 14 reference images
Try in chat$132.00 per 1M image tokens
upstream $120.00 per 1M image tokens
A price per token depends on the size and quality asked for: a larger or finer picture takes more image tokens.
“From” is the lowest of several published prices, for the smallest size or the cheapest host. Reference images may cost extra where the vendor charges for them.
Through the API
Send /v1/images/generations with your key and any id above, as OpenAI's client sends it. The docs list the parameters.