[URGENT] Ultra Subscriber: Nano Banana 2 & Pro silently failing with "Something went wrong" for 48+ hours

Gemini API
1

Hello Support and Moderation Team,

I am a Google AI Ultra subscriber and I run a children’s book publishing business. I rely on Nano Banana Pro through Flow to create illustrations for our books. My entire production workflow has been halted for over 48 hours.

The issue:
Every generation attempt using Nano Banana 2 or Nano Banana Pro returns “Something went wrong” error (“Hay aksi, bir hata oluştu” in Turkish locale). I still have significant Flow credits remaining on my Ultra subscription.

Critical diagnostic detail:
Imagen 4 generations work normally on the same account. This isolates the problem to Nano Banana models specifically — not a general account issue, not a network issue, not a credit issue.

Troubleshooting already completed:

  • Cleared browser cookies and cache for Flow
  • Switched between Nano Banana 2 and Nano Banana Pro (both fail identically)
  • Tried the Imagen 4 → switch back to Nano Banana workaround (no effect)
  • Tested with completely different, unrelated prompt types (rules out content filter)
  • Waited 48+ hours between attempt sessions (rules out transient server load)
  • Checked Library/History — failed generations are not silently completing there

What I suspect:
This appears to be either a backend sync bug affecting Nano Banana models for Ultra subscribers, or my account has been incorrectly flagged with a soft rate limit on these specific models.

Could a moderator or developer please investigate and either manually reset the limit on my account, or escalate this to the engineering team? This is causing real business impact on our publishing schedule.

Account details:

  • Subscription: Google AI Ultra
  • Region: Turkey
  • Timestamp: 23 May 2026, 21:00
  • Issue duration: 48+ hours - I can DM the account email to a moderator privately if needed. Thank you.
2

I am in the exact same boat. It seems like almost everyone who recently upgraded from the Pro tier to the new Ultra package is experiencing this exact same API block with the Nano Banana models.

It has been over 3 days now, we are officially entering the 4th day of being completely locked out of the tools we paid for. The most frustrating part is that absolutely no one from the support or engineering team is even replying, providing an update, or acknowledging that this backend sync issue is still ongoing for many of us. We upgraded to premium to improve our workflows, but instead, we are just left in the dark with zero communication. This needs an urgent fix.

3

Google will probably silently patch it at some random point, without acknowledging anything — just like they always do.

Welcome to the club. You guys have been waiting 4 days for a paid tool to work; I have been waiting 6 months for them to address critical security breaches and illegal GDPR violations.

Google support does not exist here. They actively ignore paying Ultra subscribers, just like they ignore documented cross-tenant data exposure (where deleted data reappears in other users’ accounts) and automated profiling.

If you want to see how Google handles—or rather, completely ignores—severe systemic failures, I have documented 9 major issues with video evidence and timestamps. Instead of fixing them, the moderators just lock the threads and stay silent.

Don’t hold your breath waiting for an engineering team to reply. The communication blackout is not a bug; it’s their standard operating procedure.

You can see the exact same behavior in my documented cases here. I highly recommend you take a look to understand what kind of “support” you are actually paying for: https://discuss.ai.google.dev/t/i-am-asking-for-human-comments-so-that-google-finally-responds-because-i-am-right-and-official-authorities-don-t-even-read-my-submissions/147529

And if they delete this, that only proves my point: deleting criticism is not support, it’s damage control.

4

Thank you for sharing your experience. I’m experiencing the same Nano Banana lockout, but my timeline is slightly different and may add useful diagnostic data:

I was on the Pro tier and generating hundreds of images per day successfully with Nano Banana models. Then suddenly, while still on Pro, the “Something went wrong” error started appearing on every Nano Banana 2 and Nano Banana Pro generation attempt. I upgraded to Ultra hoping it was a quota issue — but the error persists identically on Ultra. Imagen 4 continues to work normally on the same account throughout.

So for me, the trigger was NOT the tier upgrade itself. The failure started while I was still on Pro, after a period of heavy but legitimate usage. The Ultra upgrade did not resolve it.

This suggests the bug might have multiple triggers — perhaps a soft rate limit that gets stuck in the “engaged” state and isn’t released even when the user upgrades to a higher tier. If anyone from the engineering team is monitoring, the pattern across affected users seems to be:

  • Heavy legitimate usage on Nano Banana models
  • Sudden onset of “Something went wrong” errors
  • Imagen 4 unaffected (isolates issue to Nano Banana pipeline)
  • Tier upgrades do not resolve the issue

Could a moderator please escalate this? Several paid subscribers are now locked out of the exact tools they upgraded to access.

5

Same boat. This is frustrating.

6

Exact same problem here. I spoke with support for over 35 minutes and they told me they can’t help, and that its a problem on the back end. No ETA for a fix. They said to submit app feedback and that’s it. disappointing with minimal help.

7

I have the exact same symptoms as you, 100%.

8

I can confirm this 100%. You completely called it in your last sentence.

I have been actively tracking and bookmarking almost every single post and thread regarding this specific Nano Banana lockout for Ultra subscribers. As of today, nearly 2/3 of them—if not more—have been silently wiped out by the developers/moderators. This thread is one of the very few surviving places where this massive entitlement bug is still visible.

It is absolutely ridiculous and unacceptable. We are paying premium prices for an “Ultra” tier to run our professional workflows, and instead of communicating an honest ETA or pushing a real backend fix.

As you perfectly stated: deleting our bug reports is just damage control. It doesn’t fix the API, and it certainly doesn’t restore the lost days of production for paying customers. Thank you for documenting this. Let’s see how long it takes before they lock this thread too.

9

Thanks for speaking up — seriously. People need to start about this, because if we all stay silent, Google will keep us forever. They only change things.

11

Same issue, I went from Free to Ultra ($100). It worked for four images, then started giving “Oops. something went wrong.” I tried using my phone instead of my desktop. Tried using an agent, which won’t even generate an image using imgen 4. I canceled my subscription, waited a day, and resubscribed. Nothing.

Has anyone been able to get it restored?

12

Same issue here… pro was good, changed plan and no chance I get other than Imagen4 to work

13

I have had the same issue for the last two days. It was working great but yesterday started saying “Failed, oops something went wrong”. I have cleared cache, rebooted, tried the inconginto web and stilln nothing. Made sure my image was a jpeg, I can’t remember everything but I have googled for help and got no where. Help is appreciated.

14

I have the same issue. I can’t generate images after upgrade from pro to ultra by using Nano Banana 2 or pro. Videos work fine. This is so frustrating. Adding another family member to my Google One doesn’t help.

We can understand there is a bug. But we can’t accept no solution offered.

15

Aynı sorunu ben de yaşıyorum. Bugün Pro aboneliğimi Ultra aboneliğine yükselttim ve Flow’da görsel üretimim aynı hatalarla durdu.

16

Were you able to do it when you had the Pro subscription? I had no issues when I was on Pro, and my quota wasn’t full either. This whole thing started after I upgraded to Ultra just for the sake of it, and the exact same issue happened to two of my friends who switched from Pro to Ultra like me. My friend who stayed on Pro can generate images without any problems, but I think they lowered the quota and downgraded Pro as well — nanobanana was only able to make a max of 20 images with Pro, and then their quota ran out too.

17

Same issue here, Ultra account ( Turkey) While I can generate videos, I can not generate images for 1 day.

18

Quick update for everyone stuck with the Nano Banana / Flow entitlement bug:

I did some deep digging into my account and I think I might have found the root cause of this backend crash. (Just a quick disclaimer: This is my personal theory and guess based on my own account data, not an official confirmation from Google, but all the evidence heavily points to this.)

If you upgraded to Ultra recently, your account is likely stuck with a “ghost wallet” from your old billing cycle.

How to check if this is happening to you:

Take a moment to compare your balances across these two pages:

1 Go to your Google One AI Credits page. Check the credit amount and, most importantly, the expiration/refresh date.

2 Go to your Google Flow (Labs) UI and check your credit pool there.

If these two numbers do NOT match, or if your Google One page shows an expiration date anchored to your old billing cycle (before you upgraded to Ultra), you are likely experiencing this exact bug!

Here is what happened to me:

My Flow UI correctly shows my 10k Ultra pool.

However, my Google One AI page shows a random 940 credits expiring on June 18th.

I upgraded on May 20th. The 18th was the anchor date for my old Pro tier.

My guess is that Google moved us to the new dynamic percentage-based quota (% usage) but forgot to delete our legacy static wallets. When we try to generate an image in Flow, the API ignores our new Ultra entitlement and collides with that old “ghost” wallet, causing the instant crash.

If your numbers don’t match, when you contact Google One Support, don’t just say “Flow isn’t working.” Tell them specifically: “Please escalate to the backend team to manually purge the legacy ‘ghost wallet’ from my previous billing cycle and force a sync to my new Ultra entitlement.”

Hopefully, if we all point them to this database mismatch, they can investigate and deploy a global fix faster!

20

Great news everyone: The issue seems to be completely fixed on my end! :tada:

I just checked my account and I am finally able to successfully generate images using Nano Banana in Flow again without any API crashes. It looks like the engineering team might have silently pushed a fix or finally synced the backend wallets.

Go check your accounts and try generating an image—hopefully, the fix has rolled out for you all as well! Let me know if it’s working for you.

21

Great news, thank you for the update! I can confirm the same on my end — image generation with Nano Banana 2 and Nano Banana Pro is finally working again on my Ultra account, after more than 60 hours of downtime.

For context: yesterday I connected with the Google One support team via chat. They reviewed screen recordings of the errors and escalated my case to a higher tier team. I suspect the silent backend fix you mentioned is the result of that escalation — possibly affecting all of us who were stuck in the same state.

Sharing this in case it helps others who are still affected: if you are an Ultra subscriber and still seeing the “Something went wrong” or “Unusual activity detected” errors, I’d recommend reaching out to Google One support directly at one.google.com/support and requesting a manual account review. That seems to have been the trigger for the fix on my account.

Glad we are both back up and running!

22

Hello Everyone,
Thanks for the report & apologies.
We don’t believe this issue ever affected AI Studio / Gemini API. This is now confirmed as fixed in Flow.