WaveSpeedAI

Veo 429 RESOURCE_EXHAUSTED on a Paid Gemini API Project: Causes and Fixes

Veo 3.1 returns 429 RESOURCE_EXHAUSTED even though your Gemini API project is paid and the quota shows unused. The six causes, how to check each, and another way to call Veo.

By WaveSpeedAI8 min read

If a Veo request through the Gemini API fails with 429 RESOURCE_EXHAUSTED on a paid project, the cause is one of six things: the daily request quota for Veo, a spend-based limit, the key belonging to a different project than the one you upgraded, an empty prepaid balance, a usage tier that is not what you think it is, or a quota that was never provisioned correctly for the project. The first four you can fix yourself. The last two need Google.

This guide shows how to tell them apart and what to do while you wait.

Sources checked November 22, 2026: Google’s Gemini API rate limit and troubleshooting documentation, and threads on the Google AI Developers Forum linked below.

What the error looks like

429 RESOURCE_EXHAUSTED
You exceeded your current quota, please check your plan and billing details.

Developers report it on veo-3.1-generate-preview, veo-3.1-fast-generate-preview and veo-3.1-lite-generate-preview, often while text and image models on the same key keep working.

How Gemini API limits work

Four statements from Google’s rate limit documentation explain most cases:

  • “Rate limits are applied per project, not per API key.”
  • “Requests per day (RPD) quotas reset at midnight Pacific time.”
  • “Rate limits are more restricted for experimental and preview models.” The Veo 3.1 models above are preview models.
  • “Specified rate limits are not guaranteed and actual capacity may vary.”

Tiers are assigned by billing history:

TierHow you qualify
FreeAn active project
Tier 1“Set up and link an active billing account”
Tier 2“Paid $100 + 3 days from first successful payment”
Tier 3“Paid $1,000 + 30 days from first successful payment”

On top of request limits, paid tiers have spend-based limits measured over a rolling 10-minute window, and crossing one also returns 429 RESOURCE_EXHAUSTED. Video is the most expensive thing you can request, so a short burst of Veo jobs reaches a spend limit far sooner than text traffic does.

The six causes

1. You used the daily Veo quota

In the forum threads, Tier 1 projects show a Veo limit of 2 requests per minute and 10 requests per day per model. Ten clips is one afternoon of testing. Retrying does not help; the counter resets at midnight Pacific time.

2. You hit the spend limit

Several clips submitted within ten minutes can cross the Tier 1 spend limit even when the request counter still has room. Space the jobs out, or move to a higher tier.

3. The key is from another project

Limits and tiers belong to the project. If you enabled billing on one project and your code uses a key created under another, you are still on that other project’s limits. Creating a new key in the same project changes nothing.

4. The prepaid balance is empty

Google’s forum guidance on 429 errors says a depleted prepay balance also stops responses. Check the balance before you look at quotas.

5. The tier is not what you paid for

In an August 2026 thread, a developer whose project had been confirmed as Tier 2 found it back on Tier 1 a few days later, with the Veo daily limit down from 50 to 10 and generation failing. Billing support could not change it; a reply in the thread pointed to a form for reporting unexpected tier downgrades, and the issue was resolved two days later.

6. The quota shows unused and you still get 429

This is the case with no self-service fix. In a September 2026 thread, a Tier 1 developer with a positive prepaid balance saw 0/2 RPM and 0/10 RPD for every Veo model in the dashboard and still received 429 on the first request, while other Gemini calls on the same key succeeded. A second developer reported the same numbers. The only reply asked for the project id and the time of a failed request. An earlier thread from June describes the daily quota not resetting for more than 48 hours; it was closed automatically without an answer.

Check in this order

  1. Which project does the key belong to? Open the key in AI Studio and note the project. Confirm that this project, not another, has billing linked.
  2. What tier is that project on? Read it from the rate limits page for the project, not from a confirmation email.
  3. What does the Veo row show? If RPD is at its maximum, wait for midnight Pacific time. If it shows unused, continue.
  4. Is there prepaid balance, and has a spend cap been reached?
  5. Did you send several video jobs in a few minutes? Wait ten minutes and send exactly one.
  6. Still failing with unused quota? Collect the project id, the model id, the UTC time of one failed request and a screenshot of the rate limits page. Post them on the Google AI Developers Forum and, if the tier looks wrong, use the tier downgrade form mentioned in the thread above. Google states: “We offer no guarantees about increasing your rate limit.”

Handle it correctly in code

Google’s troubleshooting guide recommends exponential backoff for 429. That is right for a per-minute limit and useless for a daily one, so cap the retries and then stop.

import random
import time

def call_with_backoff(submit, max_attempts=4):
    for attempt in range(max_attempts):
        try:
            return submit()
        except Exception as err:
            if "RESOURCE_EXHAUSTED" not in str(err):
                raise
            if attempt == max_attempts - 1:
                # Per-minute limits clear within a few retries.
                # If we are still here, it is a daily or account-level limit:
                # stop retrying and switch to the second route.
                raise
            time.sleep(min(60, 2 ** attempt * 5) + random.random())

Queue video jobs instead of sending them in parallel, and keep a counter of requests per model per Pacific day so you know you are near the limit before Google tells you.

Another way to call Veo

If your product depends on Veo and a ten-request day is not enough, you need a second route that does not go through the same project quota. Google’s own alternative is Vertex AI, which has separate quotas and requires a Google Cloud setup.

WaveSpeedAI offers Veo 3.1 and Veo 3.1 Fast through its API, for text-to-video and image-to-video:

  • Available to any account with credits. There is no preview quota to request.
  • Pay per request. The price for the settings you choose is shown on the model page before you run.
  • No cold starts, and failed tasks are refunded in full.
  • One API for every model. The same two calls work for Veo, Kling, Seedance, Wan and the rest of the catalog, so a second model is a fallback you can add in minutes.
curl --request POST 'https://api.wavespeed.ai/api/v3/google/veo3.1-fast/text-to-video' \
  --header "Authorization: Bearer $WAVESPEED_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{"prompt": "A paper boat drifting down a rain-soaked street at dusk, cinematic, shallow depth of field"}'
curl --request GET "https://api.wavespeed.ai/api/v3/predictions/$PREDICTION_ID/result" \
  --header "Authorization: Bearer $WAVESPEED_API_KEY"

The first call returns data.id; the second returns data.status and, when completed, the video URL in data.outputs. Rate limits depend on your account level and are listed in the documentation.

To try it without code, open the video generator.

FAQ

Why do I get 429 RESOURCE_EXHAUSTED on Veo when my quota shows 0 used?

Either the key belongs to a different project than the one whose quota you are looking at, a spend limit or empty prepaid balance is blocking the request, or the Veo quota is not working correctly for the project. Several developers reported the last case on paid Tier 1 projects in 2026; it needs Google to look at the project.

What is the Veo rate limit on Gemini API Tier 1?

Google’s rate limit page does not list Veo. In developer reports from 2026, Tier 1 dashboards show 2 requests per minute and 10 requests per day for each Veo model. Check your own project’s rate limits page for the current numbers.

When does the Veo daily quota reset?

At midnight Pacific time, according to Google’s documentation.

Will a new API key fix the 429?

Not if it is created in the same project. Limits apply per project, not per key.

How do I get a higher Veo limit?

Reach the next tier through payment history (for Tier 2, $100 paid and three days since the first successful payment), or submit Google’s rate limit increase form. Google does not guarantee an increase.

Can I use Veo without a Gemini API quota?

Yes. Vertex AI has its own quotas, and platforms such as WaveSpeedAI offer Veo 3.1 through their own API with per-request pricing.

Share