DSQ increase request: gemini-3.1-flash-image on Vertex AI (429 after 2 sequential requests)

Gemini API
,
1

Project ID: gen-lang-client-01730506..
Model: gemini-3.1-flash-image
Endpoint: global (aiplatform.googleapis.com)
Auth: Vertex AI service account (Vertex mode, not AI Studio API key)
Billing: Enabled

Issue:
Sequential curl tests from our server (Hong Kong VPS) against Vertex AI:

  • Request 1: HTTP 200 :white_check_mark: (image returned, trafficType: ON_DEMAND)
  • Request 2: HTTP 200 :white_check_mark: (~18s later)
  • Request 3: HTTP 429 :cross_mark: RESOURCE_EXHAUSTED — “Resource has been exhausted (e.g. check quota).”
  • Request 4: HTTP 429 :cross_mark: (same error, 5s after request 3)

Quota console:
Filter gemini-3.1-flash-image shows:

  • Generate content requests with image input per minute = 34,000,000
  • Current usage: ~144 (0%)
  • Adjustable: No

Model docs say Fixed quota: Not supported for gemini-3.1-flash-image.

Use case:
Image generation SaaS. Users generate 1–4 product images per session. We already use serial requests (MaxParallel=1) with delays, but new-project DSQ floor appears too low for production.

Request:

  1. DSQ allocation increase for this project on gemini-3.1-flash-image (global)
  2. Or guidance on minimum Provisioned Throughput (GSU) for ~4 images/minute sustained

Thank you!

2

If I’m not mistaken, that’s 34M input tokens per minute, so it couldn’t have ended any other way than with a 429 error code. But I think it’s even worse than that – and it’s not the input tokens, but the requests, of which there can be a maximum of 100 per minute at Tier 1.

obraz
obraz3001×150 12.9 KB

3

Hello @boom_Jimmy ,

For issues related to the Vertex API, please contact your sales representative if you have one. Otherwise, you can find more relevant assistance on the Google Cloud Forum.