Nano Banana 2

What is Nano Banana 2?#

Nano Banana 2 is the resolution-tiered Gemini image model in the Nano Banana family, released October 2025. One resolution parameter picks 1K, 2K, or 4K and maps to three rate-card lines, so you know exactly what each output costs based on the size you asked for. It supports ten aspect ratios (1:1 through 21:9), the same set as the rest of the Nano Banana family. Reference handling lifts straight from the family with up to 20 references per call. Compared to Nano Banana Pro, this variant skips the reasoning step, so it returns faster and costs less per image.

What's new in Nano Banana 2#

Explicit resolution selection (1K, 2K, 4K)

Pick the size you actually need and pay per resolution. Useful when a campaign master goes to print at 4K and the social cuts only need 1K.

Ten aspect ratios

From standard 1:1 and 16:9 through 9:16 vertical and 21:9 ultra-wide. Covers the full range of social and web formats without a separate crop step.

Up to 20 reference images per call

Reference cap stays at 20 (10 MB each). Useful for catalog series and brand systems where the same six to eight references repeat across a campaign.

Best for#

Use cases#

Build a campaign master at 4K and downscale the same prompt to 1K for social tiles, all in one batch. Run an ecommerce catalog refresh by passing each SKU shot as a reference and prompting for a styled scene at 2K for the listing and 4K for the PDP hero. A/B test resolution against budget by generating the same prompt at 1K and 4K and comparing whether the upgrade actually shows on your typical surface.

API examples#

Call Nano Banana 2 from any language by POSTing to /v1/tasks. Full parameter docs live at docs.unifically.com/models/image/google/nano-banana-2.

curl -X POST https://api.unifically.com/v1/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "google/nano-banana-2",
    "input": {
      "prompt": "A photorealistic portrait in golden hour lighting",
      "aspect_ratio": "3:4",
      "resolution": "4K"
    }
  }'

Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or set a callback_url on the request to receive the finished result.