All insights

Image generation

ChatGPT Free Image Limit: What OpenAI Confirms and How API Costs Work

OpenAI’s Free Tier FAQ gives no exact image-generation quota. Learn what it says about limits and resets, and how to estimate image API costs from token usage.

Illustration for Chatgpt free image limit

How many images can you make with ChatGPT Free?

OpenAI’s Free Tier FAQ does not state an exact number of images. It confirms that Free users can create images, that image generation has a separate usage limit, and that ChatGPT will notify you when you reach an applicable limit. The FAQ does not give a universal image quota. (OpenAI Free Tier FAQ)

You may see 2–3 images a day quoted online. That figure is a third-party summary of user reports, not an OpenAI-confirmed quota. Another article claims a rolling 24-hour reset, but the official FAQ does not confirm that schedule. Treat neither claim as a guaranteed allowance for your account.

Budgeting takeaway: Free image generation is not a predictable production allowance. If a project needs a dependable number of images by a deadline, the FAQ does not provide a quota you can plan against.

When does the limit reset?

For image-generation limits, OpenAI says you’ll be notified when you reach an applicable limit and may need to wait until later to use the tool again. The FAQ does not specify a universal image-generation reset interval. If your limit notice gives a reset time, use that; if it doesn’t, wait until image generation is available again. Don’t assume a midnight or rolling 24-hour reset. (OpenAI Free Tier FAQ)

The FAQ lists image generation separately from file and image uploads. Uploading a picture for analysis is not the same as generating a new image, and upload limits should not be used as a generation count. (OpenAI Free Tier FAQ)

What does image generation cost through the API?

ChatGPT Free and the API are separate products and budgets. OpenAI says API usage is billed separately from a ChatGPT subscription. The GPT-Image-2.6 Flare and Sunburst model pages both describe image generation and editing from text and image inputs, and list the same token rates in U.S. dollars per 1 million tokens. (OpenAI billing FAQ; Flare model page; Sunburst model page)

API usage GPT-Image-2.6 Flare and Sunburst rate
Image output $30 per 1 million image tokens
Image input $8 per 1 million image tokens
Text input $5 per 1 million text tokens
Cached image input $2 per 1 million image tokens
Cached text input $1.25 per 1 million text tokens

These are token rates, not a fixed price per finished image. The total depends on the tokens used for image output and any image or text input. The model pages also note that API rate limits depend on your usage tier; API access is not unlimited. (Flare model page; Sunburst model page)

Illustrative arithmetic, not a typical-image estimate: If one generated image used 10,000 image-output tokens, the output component would be 10,000 ÷ 1,000,000 × $30 = $0.30. At 1,000 such outputs, that component would be $300, before image or text input and other billable usage. The 10,000-token figure is an assumption to show the calculation—not a promised or typical token count.

For a production forecast, record:

  • The model and image-generation or editing task.
  • The output settings you intend to use.
  • Expected monthly attempts, including revisions.
  • Actual input and output token usage from a representative workflow.
  • The usage-tier rate limits that apply to your API account.

Use those actual usage totals with the applicable token rates; don’t assume a flat cost per image. The model pages list rates and tier-based rate limits, but a reliable forecast depends on your own workload and measured token usage. (Flare model page; Sunburst model page)

Bottom line

  • ChatGPT Free: Image creation is available, but OpenAI’s Free Tier FAQ gives no exact image count. Check your limit notice; if it shows a reset time, follow it. Otherwise, wait until image generation is available again. (OpenAI Free Tier FAQ)
  • The often-quoted 2–3 images: Treat this as a third-party estimate, not a guaranteed quota or confirmed rolling-24-hour rule.
  • Production API: Billing is separate from ChatGPT. Flare and Sunburst use token-based rates, so estimate from actual usage and check the rate limits for your tier—not from an assumed flat price per image. (Flare model page; Sunburst model page)

Sources

ChatGPT Free Tier FAQ — OpenAI Help Center

ChatGPT Go Image Generation Limit per Day: No Official Number — AI Free API

GPT-Image-2.6 Flare Model — OpenAI API

GPT-Image-2.6 Sunburst Model — OpenAI API

Managing billing for ChatGPT and the API platform — OpenAI Help Center

Contact us