image_list plus an instruction to POST /api/v1/images/generations with model: nano-banana-2-1-edit. The model edits or combines them into one new image at 1k, 2k or 4k, priced the same no matter how many references you send.image_list plus an instruction to POST /api/v1/images/generations with model: nano-banana-2-1-edit. The model edits or combines them into one new image at 1k, 2k or 4k, priced the same no matter how many references you send.
change the background to a beach. The prompt is 1 to 2000 characters and ratio defaults to auto, letting the model choose the framing.
{"url": "https://…"} in image_list to combine a product, a model, a location and a style into one scene. Reference images are included in the price.
image_size (1k, 2k, 4k), fifteen ratio values including 1:8 and 8:1, web_search grounding and thinking_level (High or Minimal) all work in edit requests.
nano-banana-2-1-edit. For text-only generation use nano-banana-2-1.[{"url": "https://…"}].auto (default), 1:1, 16:9, 9:16, 4:3, 3:4, 2:3, 3:2, 4:5, 5:4, 21:9, 1:4, 4:1, 1:8 or 8:1.1k (default), 2k or 4k; 0.5k is not offered.High or Minimal (omit for medium, cannot be disabled); and an HTTPS URL that receives the job-detail body on every status change.image_list holds 1 reference or 9.nano-banana-2-1 and nano-banana-2-1-edit share the same table. Failed jobs are refunded automatically, and cost shows what was actually charged.{"model": "nano-banana-2-1-edit", "prompt": "change the background to a beach", "image_list": [{"url": "https://…"}]} to /api/v1/images/generations and store the img_uuid.GET /api/v1/jobs/detail or handle the callback; on success, download the image from json before link_expired_at.image_list.POST https://api.goenhance.ai/api/v1/images/generations with model: nano-banana-2-1-edit.
Between 1 and 9 in image_list. At least one is required.
No. The price depends only on image_size: 2.92 tokens ($0.0584) at 1k, 4.02 tokens ($0.0804) at 2k, 8.08 tokens ($0.1616) at 4k.
Yes, with ratio. Leave it at auto to let the model choose framing from your prompt, or pick one of the 14 fixed ratios.
One edited image per request, returned as a type: image entry in the job's json list.
The job ends with status: failed and the tokens are refunded automatically.