GPT Image 2 Cost Calculator
Estimate GPT Image 2 API spend before you launch. The calculator uses the current public GPT-image-2 token meters: text input, cached text input, image input, cached image input, and image output. Add retry waste to model billable attempts, not just successful images.
Estimate monthly API spend
Enter your expected image volume and token usage. The defaults are planning assumptions, not official per-image token counts.
| Meter | Rate | Tokens / attempt | Monthly cost |
|---|---|---|---|
| Text input | $5.00 / 1M tokens | 300 | $1.73 |
| Cached text input | $1.25 / 1M tokens | 0 | $0.00 |
| Image input | $8.00 / 1M tokens | 0 | $0.00 |
| Cached image input | $2.00 / 1M tokens | 0 | $0.00 |
| Image output | $30 / 1M tokens | 3,500 | $121 |
This calculator uses the public GPT-image-2 token rates for planning. Actual invoices depend on request parameters, tokenization, failed requests, caching behavior, taxes, and OpenAI pricing changes.
Start with a pilot batch
Run 100 real requests before launch so prompt length, output size, and failure rate are based on your own workflow.
Separate generation from editing
Image editing requests can include image input tokens, while text-to-image requests may have none.
Check limits before launch
GPT Image 2 throughput depends on your organization and project limits, including request, token, and image buckets.
How to read the estimate
GPT Image 2 pricing is token-based, so there is no single universal price per image. Resolution, prompt length, editing inputs, cached context, retries, and output token counts can all change the final invoice. Treat this as a planning calculator and reconcile it with OpenAI's pricing page and your usage dashboard before committing a customer price.
Choose the matching next guide
Pricing query
Compare free testing, direct API billing, and launch planning before picking an access path.
Pricing hubUsage-limit query
Plan throughput, response headers, 429 handling, and retry limits before launch.
Limits guideFree-access query
Start with a no-login trial when output quality matters more than production billing.
Free access