Nano Banana 2
Google's Gemini image model, the most lifelike of the ones we tested.
Official price50%off
| 1K | 2K | 4K | |
|---|---|---|---|
| Official, per image | $0.0672 | $0.1008 | $0.1512 |
| HaryAI, per image | $0.0336 | $0.0504 | $0.0756 |
US dollars. Paid from your balance; the Coding Plan is for DeepSeek. A request that fails isn't charged.
What it makes






Each prompt was run once and the first image is shown, nothing picked or retouched.
Specs
- Model name
- gemini-3.1-flash-image
- Sizes
- 1K, 2K, 4K, any common aspect ratio
- Formats
- Gemini (generateContent) or OpenAI chat (/v1/chat/completions)
Get started
Ask Claude Code or Codex to make it
Send this to your agent, with what you want in the last line. Put your key in the HARYAI_KEY environment variable first.
Make an image with Nano Banana 2 through HaryAI. Use Gemini's generateContent:
POST https://api.haryai.com/v1beta/models/gemini-3.1-flash-image:generateContent
Header: x-goog-api-key: $HARYAI_KEY
Body: {"contents": [{"role": "user", "parts": [{"text": "<my request in detail>"}]}], "generationConfig": {"responseModalities": ["IMAGE"], "imageConfig": {"aspectRatio": "16:9", "imageSize": "2K"}}}
Find the part with inlineData in candidates[0].content.parts, decode its data and save it as a PNG in this folder, then tell me the file name.
What I want: Call the API
Swap the prompt and run it.
curl https://api.haryai.com/v1beta/models/gemini-3.1-flash-image:generateContent \
-H "x-goog-api-key: $HARYAI_KEY" \
-H "Content-Type: application/json" \
-d '{
"contents": [{"role": "user", "parts": [{"text": "A cup of green tea on a wooden table by a window, morning light"}]}],
"generationConfig": {"responseModalities": ["IMAGE"], "imageConfig": {"aspectRatio": "16:9", "imageSize": "2K"}}
}'Questions
Is it the real Nano Banana 2?
Yes. Every request goes to Google's gemini-3.1-flash-image, and each reply carries Google's own usage figures.
Can I use the OpenAI images endpoint?
Not for this model. Use Gemini's generateContent, or OpenAI chat completions, where the image comes back in the message.
Do I need a separate key?
No. The key you use for DeepSeek and Seedance makes images too, from the same balance.