An Internal Error Has Occurred — Gemini | Clear It Fast

You’re mid-prompt, Gemini is thinking, and then the page tosses out the same blunt line. “An internal error has occurred.” It’s annoying, but it’s also a clue. This message is a generic catch-all for a small set of patterns, a stale session, a browser problem, a request that’s too heavy, or a hiccup on Google’s side. If you searched for an internal error has occurred — gemini, this checklist will help.

This page walks you through checks that you can do in minutes, then deeper steps that solve the stubborn cases. It also calls out the spots where waiting is the right move, so you don’t burn time changing settings that won’t matter.

What The Gemini Internal Error Usually Means

Gemini runs as a web app and as mobile apps. Under the hood, each prompt becomes a request that has to pass sign-in, policy checks, model capacity, and then content generation. When any part of that chain breaks, the interface may show the same plain error line.

The tricky part is that the same wording can show up for different products. You might see it in the Gemini web chat, in Google AI Studio, or in an app that uses the Gemini API. The steps below still apply, because they target the shared causes, session state, request size, network path, and service health.

Three common triggers

• Stale sign-in state — Cookies or tokens don’t match what the service expects, so the request fails before it reaches the model.
• Heavy input or files — Long conversations, large pasted text, images, or many attachments can push limits or trip a backend error.
• Capacity spikes — During busy periods, some requests fail even with a good prompt and a clean setup.

If your error appears after you pasted a huge chunk of text, added several files, or turned on a feature like code execution, start with request size. If it appears on the first prompt of a new chat, start with sign-in and browser steps.

An Internal Error Has Occurred — Gemini On Web And Mobile

Start with the light fixes. They solve a lot of cases because this error often comes from a temporary mismatch between the page and your account session.

1. Refresh the page — Close the tab, open Gemini again, then send the prompt once more.
2. Start a new chat — Open a fresh thread and send a shorter version of the same request.
3. Sign out and sign back in — Log out of your Google account in the browser, then log back in and retry.
4. Try a private window — Use an incognito/private window to test with zero extensions and a clean cookie jar.
5. Switch networks — Move from Wi-Fi to mobile data, or the other way around, to rule out a network block or flaky DNS.

If one of these steps works, you’ve learned something. A private window working points to cookies, extensions, or cached site data.

When the chat itself is the problem

Sometimes the thread gets “too loaded.” That can mean the conversation is long, the model is juggling many uploaded items, or the tab has been open for hours with several retries. The easiest test is to shrink what you send.

• Paste less at once — Split long text into smaller chunks and send them across multiple turns.
• Remove files — Detach files one by one and retry, since one file can be the trigger.
• Use plain text — If you’re pasting rich text from a doc, paste into Notepad first, then paste into Gemini.
• Ask for a summary first — Send a short prompt that asks Gemini to confirm what it can see, then build up from there.

These steps feel simple, yet they map to what Google documents for Gemini API “INTERNAL” errors. Shrinking the input context and trying again can clear a backend failure.

Browser And Account Fixes That Change The Outcome

If the error keeps returning in the same browser, treat it like a state problem. You want to clear the pieces that can get out of sync, site storage, cached scripts, cookies, and extensions that alter requests.

Clean site data without nuking everything

Clearing all browser data can be a pain, so start narrow. Remove stored data for Gemini and related Google domains, then restart the browser. This forces a new session and reloads the app code.

1. Clear site storage — In your browser settings, clear cookies and site data for Gemini and Google sign-in domains.
2. Hard reload — Reload the page with a hard refresh so scripts and assets reload from the network.
3. Restart the browser — Fully quit the browser, reopen it, then test again.

Extension and privacy settings checks

Ad blockers, script blockers, privacy tools, and some antivirus web shields can break the request flow. You don’t need to delete them.

• Disable extensions — Turn off all extensions, test Gemini, then re-enable them one at a time.
• Allow third-party cookies for Gemini — If your browser blocks them globally, add an allow rule for Gemini and Google sign-in.
• Turn off strict tracking protection — Test with a standard mode, then switch it back after you confirm the cause.

If the error only happens on one Google account, open the same browser profile, sign into a second account, and run a test prompt. If the second account works, the issue may be tied to account state, billing, age verification, or a policy flag. In that case, switching devices won’t always help.

Network And Device Checks That Stop Hidden Blocks

When Gemini works on mobile data but fails on your home or office Wi-Fi, your network path is the likely culprit. Some routers filter traffic, some DNS resolvers are flaky, and some workplaces block parts of Google services.

Fast network sanity tests

• Restart the router — Power cycle the router and modem, then try Gemini again.
• Change DNS — Use a reliable public DNS resolver in your device or router settings and retest.
• Turn off VPN or proxy — Disable VPN/proxy settings and test, since some exit nodes trigger rate limits.
• Check device time — Make sure your device clock is set to automatic time and date.

On mobile, app updates matter. An older build can break after a backend change. Update the Gemini app, update Chrome/WebView on Android, then restart the phone.

1. Update the app — Install the latest Gemini app update from your app store.
2. Update system components — On Android, update Chrome and Android System WebView.
3. Restart the device — Reboot to clear stuck network stacks and cached web components.

Internal Error Has Occurred In Gemini Requests After Long Prompts

Long prompts are the most common self-inflicted trigger. A long chat history plus a fresh wall of text can push the request beyond safe limits. Even when the product tries to trim context, the backend can still return a 500 “INTERNAL” style error.

Google’s Gemini API troubleshooting guide notes that an INTERNAL error can happen on Google’s side and also calls out overly long input context as a common cause. Cutting the request down is the cleanest test.

Ways to shrink the request without losing the goal

• Recap the thread — Ask Gemini for a 10-bullet recap of the chat so far, then start a new thread with that recap.
• Send only the needed excerpt — Trim pasted text to the part that the model must read to answer.
• Move files to a smaller set — If you attached many items, try one at a time and keep the smallest set that still works.
• Swap to a lighter model — If your interface offers it, try a faster model for the heavy lift, then move back.

When the error happens after uploading files, treat it like a file problem until proven otherwise. Upload a single file, test, then add the next one. If the error returns, you’ve found the trigger item.

AI Studio And Gemini API Paths

If you’re using Google AI Studio or calling the Gemini API from code, the same message can be paired with a 500 status. In those cases, you can use the tooling around the request to narrow the cause faster.

Where You See It	Likely Cause	Try First
AI Studio after big file import	One file triggers parsing or context limits	Import in smaller batches
API returns 500 INTERNAL	Backend error or input context too long	Cut input and retry
Error spikes for many users	Service incident or capacity issue	Check status and wait

Two pages are worth bookmarking if you work in AI Studio or with the API. The AI Studio status page can confirm outages. The Gemini API troubleshooting guide explains common error codes and points out that shrinking input context and retrying can help with INTERNAL errors.

• Check AI Studio status — Visit Google AI Studio Status to see active incidents.
• Read API troubleshooting — Use Gemini API Troubleshooting to map status codes to next steps.

Pair retries with a guard that trims context once it hits your chosen ceiling. That combo saves time, keeps your logs cleaner, and cuts the odds of looping on the same failing request.

When To Report The Error And What To Capture

If the error persists across a private window, a second network, and a shorter prompt, it’s time to collect details and send feedback. If an internal error has occurred — gemini shows up on every try, include that detail. You want to make it easy for an engineer to replay what happened without seeing your private content.

• Record the time — Note the local time and your time zone when the error happened.
• Write the surface — Note whether it happened on web chat, mobile app, AI Studio, or the API.
• Save the prompt shape — Describe what you sent without pasting sensitive text, like “2 images + 3,000 words + code execution on.”
• Capture any IDs — If the UI shows a request ID or error code, copy it.
• List what you tried — Mention the steps you already ran so your report isn’t bounced back with generic advice.

Also protect your work. If you’re running long sessions, keep a local copy of prompts and notes. If a single chat thread matters, export or copy the parts you can’t easily rebuild before you keep testing, since repeated failures can lock you in a loop where the thread won’t load.

Last, if the error appears for everyone at once, don’t fight it. Check the status pages, take a break, then retry later with the same short prompt.