I’m encountering a persistent 429 error on my paid Google AI Studio API account, even though my usage dashboard clearly shows I’m nowhere near any limits.
Error Response:
json
{
"error": {
"code": 429,
"message": "Resource has been exhausted (e.g. check quota).",
"status": "RESOURCE_EXHAUSTED"
}
}
The Problem: As you can see from the numbers above and the attached screenshot, I’m well within all limits across all models. The highest usage is only 36% of TPM on gemini-2.5-flash, yet I’m getting RESOURCE_EXHAUSTED errors.
The graphs in my dashboard show current usage at 0 for all three metrics (RPM, TPM, RPD), indicating no active rate limiting should be happening.
Questions:
Is there a hidden quota or limit not displayed in the dashboard?
Could this be a bug with quota tracking or enforcement?
Has anyone else experienced this discrepancy between dashboard limits and API errors?
Are there regional or other restrictions that might not show in the main quota view?
This is blocking my production application, and I’d greatly appreciate any insights or solutions.
You will have to buy provisioned throughput units. We also ran into same problems with gemini 1.5 flash in the past and we were asked to buy provisioned throughput units
Hi, I’m facing the same issue: a 429 Resource Exhausted error, even though I’m using a paid API key for the Gemini 2.5 Flash model. This issue is persistent.
There is also another problem caused by these 429 errors. After an error occurs, when I submit a new query for the agent to perform a different task, the agent continues using the previous incomplete workflow along with the query that initiated it. As a result, the agent tries to resume or rerun the earlier workflow instead of focusing on the new query.
Is there a way to overcome this issue by clearing or removing the incomplete workflow from the agent’s context, so that each new query starts as a fresh workflow instead of continuing from the previous one?
I am encountering this issue with the Gemini 3 Flash model as well. I am on a paid Tier 1 account and am nowhere near my usage limits.
This started happening on my free account, but the problem persists even after upgrading to paid. I recently burned through a significant amount of credits due to these errors. I have since fixed the retry logic on my end to prevent wasted credits, but I am still receiving 429 errors.
Error: model unreachable: RESOURCE_EXHAUSTED (code 429): Resource has been exhausted (e.g. check quota).: RESOURCE_EXHAUSTED (code 429): Resource has been exhausted (e.g. check quota).: model api cannot be reached
I’ve tried multiple networks, which doesn’t seem to be the issue.
I’m also running into this 429 / RESOURCE_EXHAUSTED issue on a paid setup.
Account / setup
Google Workspace domain: sonicrocket.be
Google AI / Gemini project number:
I have a Workspace with Gemini Ultra / AI Ultra subscription.
Issue
I repeatedly get:
“Error: model unreachable: RESOURCE_EXHAUSTED (code 429): Resource has been exhausted (e.g. check quota).: RESOURCE_EXHAUSTED (code 429): Resource has been exhausted (e.g. check quota).: model api cannot be reached”
This happens even when I’m nowhere near any visible quota limits. It’s bad enough that I can realistically use Gemini only about half of the time; the other half is blocked by this error.
What I’ve checked
Usage in the quota dashboards is well below limits.
Tried from different networks and environments, same result.
This looks like an account‑ or project‑level backend issue, not a local/network problem.
Request
Could someone from Google please:
Check project **** and my Workspace domain sonicrocket.be for any stuck/hidden quota or capacity issue that could cause these persistent 429 errors, and
Confirm whether this matches the ongoing 429 “RESOURCE_EXHAUSTED” problem that other paid users are seeing here?
Happy to provide more trajectory IDs, timestamps, or screenshots via DM if needed.
@chunduriv we run Gemini in a production environment with over 1800 employees relying on Gemini to perform certain tasks. I am on paid Tier 1 and my utilization for this month is under $2.00 but I am getting resources exausted. We are using gemini-3-pro-preview.
@chunduriv Hi, I’m also experiencing the exact same 429 RESOURCE_EXHAUSTED issue on my paid AI Studio project.
Project ID: gen-lang-client-
Models affected: Gemini 3 Pro (TPM 1.23M / 1M), Gemini 3 Flash (TPM 1.87M / 1M)
Haven’t used these models for several days, but still getting persistent 429 errors
Other models (Gemini 2.5 Flash) work fine with no issues
Usage dashboard clearly shows:
Gemini 3 Pro/Flash: TPM exceeded on 28-day window
All other models: Low usage, no issues
Current graphs show 0% usage on all metrics
This is affecting my development workflow. Would appreciate if you could investigate my project. Happy to provide more details or trajectory IDs as needed.
I [gen-lang-client] am experiencing a similar concern
I have made about 70 requests today and only had 37.14% success; failures were due to rate-limit - 429 Error (Too Many Requests; refer traceId 3062b4dc0f59b01994405d5696ae661b)
I am using Gemini 2.0 Flash
I have done 4/2k RPM, 28.2K/4M TPM and 69/Unlimited RPD
