Skip to content

GPT Image 1

Overview

GPT Image 1 is OpenAI’s natively multimodal image generation model. OpenAI’s developer documentation describes it as a model that accepts both text and image inputs and produces image outputs, with stronger instruction following, contextual awareness, and editing fidelity than the older DALL·E line. That makes it the most natural OpenAI model when the workflow needs generation and editing to live on the same API surface.

API base URL

https://api.openai.com/v1/images/generations

Example payload

{
  "model": "gpt-image-1",
  "prompt": "Create a premium product hero shot on a neutral studio background",
  "n": 1,
  "size": "1024x1024",
  "quality": "high",
  "background": "auto",
  "output_format": "png"
}

Our Current Pricing

GPT Image 1 is the first image model whose primary billing path is genuinely usage-based. The implemented logic reads token counts from the API response and calculates cost from those numbers.

input_token_price_usd  = 0.00001
output_token_price_usd = 0.00004
cost_usd = (input_tokens * 0.00001) + (output_tokens * 0.00004)

The activity uses result.usage.input_tokens and result.usage.output_tokens, with missing values treated as zero. Final rounding is to 6 decimal places.

There is an older helper in the activity with size-and-quality estimates, but for gpt-image-1 the repo explicitly prefers the token-based branch. That means the parameters that may influence price are not just obvious creative controls such as size, background, format, mask, or fidelity. They matter indirectly because they change the eventual input and output token counts reported by the provider.

The right way to bill this model is to persist the raw provider usage object in full, then compute the charge from usage rather than trying to infer it from request knobs. If usage is unexpectedly absent, you can either mark the request as unbillable until reconciliation or fall back to a temporary estimate, but that fallback should be flagged clearly because it is not the primary contract implemented in code.

Sources