AI image models
Every tool on this site runs on one of the models below, and most of them let you switch between all 18 from a dropdown. You are not locked into whichever one we happened to pick. The price is shown next to the model name before you generate, and one credit balance covers all of them.
Model by model
- Google
Nano Banana 2.1
Google's newest Nano Banana. 1K to 4K and 14 aspect ratios down to 1:8; Google says it renders text better than 2.
- Google
Nano Banana
Gemini 2.5 Flash. Follows an editing instruction most literally, and reads several reference images at once.
- Black Forest Labs
FLUX.1 Kontext
The fastest editing model here, and the default behind Image to Image and the seven redraw tools.
- ByteDance
Seedream 5
ByteDance's model, in a cheap Lite tier and a 2K Pro tier. The most photorealistic option here.
- OpenAI
GPT Image
OpenAI's models in four tiers, up to 4K. The strongest here when the picture has to contain readable text.
- Ideogram
Ideogram
Ideogram V3, built around readable text inside the image: posters, logos, social graphics. From 5 credits.
Coming soon: GPT Image 2.6 (OpenAI), FLUX 3 (Black Forest Labs). Not available yet; each page says what we know and what to use meanwhile.
Every tier, side by side
Credits are per image. A tier is a setting on the model, not a different model: GPT Image 2 at 2K and at 4K are the same engine at two resolutions. The 3 4K tiers are on paid plans.
| Tier | What it is good at | Credits |
|---|---|---|
| Z-Image Turbo | Alibaba · cheapest and fastest, text-to-image only | 1 |
| Seedream 5 Pro · 1K | ByteDance · same model as 2K at half the price | 9 |
| GPT Image 2.5 Flare · 2K | OpenAI · same price as Sunburst, tuned for speed and volume | 13 |
| Qwen Image 3 · 2K | Alibaba · strong at readable text inside the image | 6 |
| Nano Banana | Google Gemini 2.5 Flash · fastest, cheapest | 5 |
| Nano Banana 2.1 · 1K | Google · newest Nano Banana, sharper text, banner ratios | 10 |
| Nano Banana 2.1 · 2K | Google · same model at 2K | 15 |
| Nano Banana 2.1 · 4KPaid plans | Google · for print or heavy cropping | 23 |
| Ideogram V3 · Turbo | Ideogram · text inside the image, fastest tier | 5 |
| Ideogram V3 · Balanced | Ideogram · text inside the image, posters and logos | 9 |
| Ideogram V3 · Quality | Ideogram · text inside the image, highest detail | 13 |
| FLUX.1 Kontext | Black Forest Labs · fastest, about 5s per image | 5 |
| Seedream 5 Lite | ByteDance · good at Chinese text and posters | 7 |
| Seedream 5 Pro · 2K | ByteDance · most photorealistic, highest detail | 18 |
| GPT Image 2 · 2K | OpenAI · good with text in images, half the price of 2.5 | 7 |
| GPT Image 2 · 4KPaid plans | OpenAI · for print or heavy cropping | 10 |
| GPT Image 2.5 · 2K | OpenAI · newest, better instruction following | 13 |
| GPT Image 2.5 · 4KPaid plans | OpenAI · newest, highest resolution | 20 |
How to choose
- Editing a photo and want it back quickly
- FLUX.1 Kontext, 5 credits. Our edit runs came back in 18 to 27 seconds, the fastest here.
- A long instruction the model keeps ignoring
- Nano Banana at the same price, or GPT Image 2.5 if you can spend more. Both hold on to more of a complex prompt.
- It has to pass as a photograph
- Seedream 5 Pro at 18 credits, rendering at 2K.
- The picture contains words someone has to read
- GPT Image first, Seedream second. Put the exact wording in quotes and keep it short.
- The tool takes two images, like Virtual Try-On
- Anything except FLUX.1 Kontext. Its input takes a single image, so it would never see the second one and would invent it instead. That dropdown leaves it out on purpose.
Try them against each other
Same prompt, two models, side by side. The 10 free credits you get on sign-in cover 10 runs at the cheapest tier.
Open the generator