Nano Banana Pro vs Nano Banana 2: which id to send on Sume

On Sume the two models take the same request: google/nano-banana-pro and google/nano-banana-2 both accept 512, 1K, 2K, 4K resolution tiers and up to 10 input_references. The differences are in the catalog: Nano Banana 2 lists 4:1, 1:4, 8:1, 1:8 extra aspect ratios, and the two price differently by tier.

This reads the Image API docs and Sume's pricing code on 2026-09-29. Google's naming for its own Nano Banana models is on its image generation page.

What is different in the request?

Nothing but the model value. The catalog descriptors decide what each accepts, and a parameter a model does not list is rejected with 400 unsupported_parameter. Read GET /v1/images/models instead of assuming.

Does either one have separate character reference slots?

Google's own docs split references by role on its API: up to 4 character images on Gemini 3.1 Flash Image and up to 5 on Gemini 3 Pro Image. Sume's docs list a single input_references array for both ids and do not describe character slots, so on Sume both are simply reference images you describe in the prompt.

Which ratio do I use for a 4:5 Instagram post?

Both list 4:5. Send aspect_ratio: "4:5"; Sume's docs note Nano Banana Pro renders about 928×1152 at 1K, and that exact 1080×1350 is a post-step through job target_pixels, not a native output.

How do I switch between them?

Change the string, and keep every other field the two ids share.

curl -X POST https://api.sume.com/v1/images \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/nano-banana-pro",
    "prompt": "Product photo of a leather wallet on slate",
    "resolution": "2K",
    "aspect_ratio": "4:5"
  }'