AI image generator API cost: price per image, model by model

To estimate what an AI image generator API will cost, multiply the model's price per image by the number of images you plan to generate, then check whether the resolution or quality you pick changes that price. On Sume, flat rates run from $0.025 to $0.09375 per image, Nano Banana prices rise with the resolution tier up to $0.375 for a 4K Nano Banana Pro image, ChatGPT Image prices move with quality and size, and a failed generation costs nothing.

The prices below are computed from the pricing code behind GET /v1/images/models, and the billing rules come from the Image API docs, both read on 2026-09-27. Prices change, so read the catalog before you budget.

How much does each image model cost per image?

Each price is the provider's list rate × 1.25, so Sume's margin is already in it:

Why does one image request cost more than another?

• n multiplies the price: four images cost four times one.
• Nano Banana 2 and Nano Banana Pro price by resolution tier. The 1K rate is the default and the one the catalog line shows.
• ChatGPT Image models are priced by tokens, so quality and size move the price. On ChatGPT Image 2.5, the prompt and any reference images add estimated input tokens, auto quality reserves max, and auto size or a named preset without a verified pixel mapping reserves the output-token upper bound. Send an explicit quality and an aspect_ratio such as "1:1", or a custom size.
• At these rates, 1,000 images cost $25.00 on x-ai/grok-image and $187.50 on Nano Banana Pro at 1K, before the agent fee.

What exactly gets billed?

Image generation billing is all-or-nothing. A completed generation is billed in full at the endpoint price. A failed or cancelled one is not billed, and a failed request returns 502.

Catalog prices include Sume's margin, plus a 5.5% agent fee by default, explained in the Sume agent fee. Every 200 response carries usage.cost, the USD amount billed to your wallet, and its token counts are always 0 in v1. Sume reserves the estimate when it accepts a request and releases or refunds it if the generation fails.

How do I check the current price before I call?

Read the model's endpoint record. Its pricing lines are billable lines of the form output_image per image, with cost_usd per image. sume/auto is not listed in the catalog; the usage.cost on its response shows what the request billed. API pricing lists the Image Router as pass-through per model, with per-model rates in the catalog.

curl "https://api.sume.com/v1/images/models/bytedance-seed/seedream-4.5/endpoints" \
  -H "Authorization: Bearer $SUME_API_KEY"