My web app lets users record 20-30 second voice clips with Gemini 2.5 Flash Preview TTS. We’re still in private beta, but usage is high enough that we burn through the Tier 1 cap (100 requests/day, 10 RPM) by early afternoon, leaving the rest of the day full of 429s.
Steps I’ve taken so far
Submitted the quota-increase form with traffic logs and screenshots – no response after a week.
Live caching + batching: helps but still hits the hard daily ceiling.
Looked into spinning up extra projects, but that feels brittle and possibly against ToS.
The catch-22 with promo credits
Google generously gave our account $300 in promotional Cloud credits, which is great—except those credits don’t count toward the $250 “billable spend” threshold for Tier 2. In practice that means:
I must burn the promo credits first (weeks),
then run up $250 of out-of-pocket spend,
then wait 30 days for an auto-upgrade.
That could be two months of throttled users and broken sessions—pretty rough for an early-stage product.
Questions / plea for help
Can Google grant a temporary quota bump (e.g. 500 RPD) while we prove demand?
Has anyone else succeeded in upgrading early even with promo credits still active?
Are preview TTS models absolutely fixed at 100 RPD, or are exceptions possible?
Any best-practice suggestions beyond “make more projects”?
Happy to DM project ID if a Googler can take a look.
We love Gemini’s TTS quality and want to pay for higher usage—it just feels unfair that promo credits slow down the very upgrade path designed for growth.
Hi @Krish_Varnakavi1 – thanks for escalating this and for the best-practices links!
I’ve reviewed the quota-management and caching guides. They make sense, but in our particular workflow they don’t move the needle:
Recommendation
Why it’s tough for our use case
Batching / queueing
Users generate 20–30 s voice clips from a browser UI and expect audio back almost instantly (chat-style UX). Adding a queue with deliberate delays would break that real-time feel.
Content caching
Each text prompt is unique and rarely repeated, so hit-rate would be near zero. (We did test hashing prompts over 48 h: only one duplicate out of ~400 calls.)
So even with perfect throttling we still hard-stop at the 100 requests/day (10 RPM) Tier 1 ceiling by late morning.
What would really help
A quota increase while we validate demand in private beta.
Clarification on the promo-credit loophole: because promo spend doesn’t count toward the $250 threshold, we can’t trigger the 30-day Tier 2 timer—even though we’re ready to pay once credits are gone. Is that expected behavior?
Any alternative path (e.g., shorter review cycle for early adopters) we could apply for.
Happy to DM project ID, support-case number, and request logs if that speeds things up.
im in the same boat…i just spun up 3 servers at 1.30 per hour…and ran up a $250 bill
now im waiting …by july 14th should have my upgrade …gotta do what you gotta do
google is a big company, wont be any personal attention to matters like this
Hey Krish, i haven’t heard from anyone from the team. i desperately need tier upgrade. Meanwhile. could you please confirm if my promo credit usage adds toward the spending threshold? or i need to spend 300(promo)+250$ for an upgrade?
It took some digging to get to the right team to answer some of your questions..
It is indeed correct that promotional credits do not count toward the $250 “billable spend” threshold for Tier 2 upgrades. As a result, you’ll need to spend the required amount in actual charges before being eligible for the upgrade.
Unfortunately, there is no temporary bump feature for Tier 1 limits, so the only solution would be to spent the actual amount and your account will automatically qualify for the upgrade after 30 days.
I understand your frustration and pressure about your private beta release.. I am personally handling your situation and still trying to find solution for you.. Unfortunately at this point, this is what I have for you..
As you initiated this thread 15 days ago, I encourage you to quickly spend $250 and let the 30 days clock start while I explore other options..
I also recommend you to raise a ticket with Billing Team as this department handles such issues..
Please DM me your request logs, billing ticket information as well..
Hi @Krish_Varnakavi1 – thanks a ton for tracking this down and confirming the promo‑credit rule. I really appreciate you chasing the right team for answers.
Where that leaves us
Our Limit = ~100 Flash‑TTS calls / day
Cost = 1$ /day** in real spend
At that burn‑rate it would take 11 months to reach $250, then another 30 days before Tier 2 kicks in.
Switching to the Pro TTS model would eat money faster but also hits the stricter 50 RPD cap, so we’d still stall out long before $250.
Spending $250 up‑front myself just to start the 30‑day clock isn’t feasible for a self‑funded indie project—especially when the current rate limit blocks the user growth needed to justify that spend.
What I’m doing in the meantime
I’ve moved traffic to a third‑party reseller API. It’s more expensive per token, but at least there’s no hard 100 RPD wall, so our beta can keep moving. Long‑term I’d rather stay on Google’s native API for stability and price transparency.
Is there any path like…
A pilot / startup waiver that lowers the Tier 2 threshold for low‑ARPU workloads?
A temporary quota increase tied to usage growth (e.g. 100 → 500 RPD while we prove demand)?
Even a pay‑as‑you‑go surcharge (say, +20 %) in exchange for higher caps would be fine—we just need the ceiling lifted.
Happy to DM request logs, billing details, and reseller invoices if that helps frame the business case. I’m also happy to open a Billing ticket, but my lifetime spend is so low I was not able to as spends are still “below threshold.”.
Thanks again for sticking with this. If the team has any early‑stage builder programme—or even a wait‑list—I’m all ears.
I will consider your feedback and share it with the internal team.. As of now, the current tier upgrade conditions are configured based on various user level data and to prevent abuse of services…
I will let you know if anything changes in the near future..
The basic failing is that this seems to be modeled after other providers “tier” system like OpenAI or Anthropic.
However, in those cases, using with AI calls is not what is measured, and making API usage is not necessary. Only the amount paid is considered (along with some maturation time).
This current Google system is flawed. It seems the approach necessary is to allow loading money to an account balance in the same way that promo credits are added - but count them immediately. Also, count any usage as “usage”.
Otherwise, it is simply impossible to transition a production application to Google AI Studio API. One would have to set fire to $500 first.
(what other Google product does anything like this? It is as nonsensical as were I to need to make $100 in international calls with Google Voice and receive 1000 calls before they’ll increase the number of daily voicemails I can be left without the caller receiving an error.)
