An OpenAI model for images and editing that shines when the frame needs text: a price, a name, a button. Three quality levels let you draft for a cent and render the final version at high quality.
| Developer | OpenAI |
|---|---|
| Modes | text to image, photo editing |
| Quality | low, medium, high |
| Ratios | 1:1, 3:2, 2:3, 4:3, 3:4, 16:9, 9:16 |
| Content filter | strict |
| Quality | 9:16 and 16:9 | 1:1 |
|---|---|---|
| Low | $0.01 | $0.01 |
| Medium | $0.01 | $0.02 |
| High | $0.04 | $0.07 |
Quoted captions end up in the picture almost without mistakes — handy for banners and covers.
At high quality 9:16 and 16:9 cost $0.04, a 1:1 square $0.07.
The model is filtered and refuses explicit or borderline prompts. A refusal is not charged.
Changes details, background and text on an uploaded picture and keeps the rest.
At low quality the price is the same. At high quality 2.5 is much cheaper: $0.04 against $0.22 for a vertical frame.
GPT has a strict filter. Describe people as adults with an age and drop ambiguous words. Refusals are free.
Low to find the idea, high for the final with text and small details.