Nano Banana batch API: Gemini's 24 h batch vs Sume async jobs

Google lets every Gemini image capability run as a Batch API job: higher rate limits in exchange for up to 24 hours of turnaround. The Sume docs describe no equivalent batch tier or discount for images. To make many Nano Banana images on Sume, submit separate POST /v1/images requests with mode: "async" or mode: "webhook" and read each job's result.

Google's batch note is from its image generation page; Sume's modes from Image generation and Jobs and results, read 2026-10-01.

What does Google's Batch API offer?

The page says all image generation capabilities can also run as batch jobs, "ideal if you need to generate many images", with higher rate limits and a turnaround of up to 24 hours. It states no price for batch on that page, so check Google's pricing before assuming a discount.

What are the Sume equivalents?

Sume's image route blocks for up to 30 seconds by default and returns 200 with the images. If generation outlasts that budget, or you send mode: "async", or mode: "webhook" with a webhook_url, you get 202 and a job envelope. Poll GET /v1/jobs/{id}/status and fetch GET /v1/jobs/{id}/result.

How should I run a large set of images?

Submit each image as its own async job, keep the returned job ids, and collect results as they finish; a webhook saves you from polling. Mind your plan's rate limit: a 429 means back off and retry after retry-after. For concurrency patterns see batch image generation concurrency limits.

Can one request return more than one image?

Yes, with n, but the ceiling is per model. The code comment for the shared range says every other image model takes n as a 1 to 4 range; the docs say to read the n descriptor from the catalog rather than assume.