# VELIN

> VELIN (https://72agi.com) is a pay-per-image AI image generation API. One API key, one asynchronous REST endpoint, seven models including Nano Banana Pro, Nano Banana 2 and GPT Image 2 / 2.5. Standard models are a flat ¥0.25 (about $0.037 USD) per image at 1K, 2K and 4K; the Pro+ and 2+ variants are ¥0.35 (about $0.052). Failed generations are not charged. Up to 14 reference images per request. Capacity up to 100,000 images per day; volume pricing on request. It is not OpenAI-compatible: it has its own small REST API.

## Facts
- Price: ¥0.25 (about $0.037) per image for Nano Banana Pro, Nano Banana 2, GPT Image 2, GPT Image 2.5 Flare and GPT Image 2.5 Sunburst; ¥0.35 (about $0.052) for Nano Banana Pro+ and Nano Banana 2+. Same price at 1K, 2K and 4K. The low price comes from stable bulk supply.
- Billing: prepaid balance held in CNY, deducted per image at submit time and refunded automatically if the task fails. Top-up with USDT. Card payments are coming soon. No subscription. No automatic free credits; a free test is available on request via Telegram @ayan8866e.
- Typical generation time: about 45-100 seconds (4K and many reference images take longer).
- Limits: 8 single-image generations per minute per account by default (HTTP 429 above that); up to 14 reference images, 8 MB each; generated images are deleted after 10 hours by default. Prompts and generated images are deleted after 10 hours; reference images are deleted as soon as the job ends.
- Contact / volume pricing / free test: Telegram https://t.me/ayan8866e

## API (base URL https://72agi.com)
- Auth: header `Authorization: Bearer <API_KEY>`. Get the key from https://72agi.com/me.html after signing in. Never put the key in a URL. Call from a backend (no CORS headers).
- `GET /api/models` (no auth): list of models with ids and prices.
- `POST /api/generate` (multipart/form-data): fields `prompt`, `model`, `size` (aspect ratio such as 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3), `resolution` (1K, 2K or 4K), optional repeated `refs` (image files, max 14). Returns 202 `{"id": "...", "status": "queued"}`. 402 = not enough credits (body has `need` and `have`), 429 = rate limited.
- `GET /api/task/{id}`: poll every 3-5 s. `status` is queued, processing, succeeded or failed. On success: `{"status": "succeeded", "url": "/outputs/<id>.png", "price": 0.25}`; the image is at https://72agi.com + url. On failure the task is not charged and `error` explains why.
- `GET /api/credits`: balance in CNY.
- Model ids: nano-banana-pro, nano-banana-2, nano-banana-pro-plus, nano-banana-2-plus, gpt-image-2, gpt-image-2.5-flare, gpt-image-2.5-sunburst.

Example:
```
curl -X POST https://72agi.com/api/generate \
  -H "Authorization: Bearer $VELIN_API_KEY" \
  -F "prompt=a quiet room, cream light" -F "model=gpt-image-2.5-flare" \
  -F "size=16:9" -F "resolution=1K"
curl https://72agi.com/api/task/<id> -H "Authorization: Bearer $VELIN_API_KEY"
```

## Docs and guides
- [API docs](https://72agi.com/docs.html): authentication, models, generate, polling, reference images, limits
- [Nano Banana Pro / Nano Banana 2 API pricing](https://72agi.com/nano-banana-api-pricing.html): flat price table, comparison with Google's published Gemini API list prices, cost at 1k / 10k / 100k images
- [Nano Banana API price comparison](https://72agi.com/nano-banana-api-price-comparison.html): per-image prices at 1K / 2K / 4K from each provider's own pricing page (read 2026-09-29), VELIN included and disclosed as operated by the page publisher
- [GPT Image 2 / 2.5 API](https://72agi.com/gpt-image-2-api.html): models, flat price, code snippets, reference images
- [SillyTavern image generation](https://72agi.com/sillytavern-image-api.html): local adapter that lets SillyTavern use VELIN
- [n8n image workflow](https://72agi.com/n8n-image-api.html): importable n8n workflow
- [Billing / top-up](https://72agi.com/billing.html)
- [GitHub kit](https://github.com/velin-api/velin-image-api): Python, Node, curl and n8n examples