Is Gemini down right now?
Google doesn't keep one Gemini status page. It keeps three, and which one to open depends on how you reach Gemini. Pick your row, open the page, and read the Gemini entry. Don't trust a page that isn't yours. Gemini is down only if the page for your route says so.

| How you use Gemini | Page to open | What it covers |
|---|---|---|
| gemini.google.com and the Gemini app, personal or Workspace account | Google Workspace Status Dashboard | Rows named Gemini and Gemini Notebook, among 37 products |
| An API key from AI Studio, or the AI Studio website | Google AI Studio and the Gemini API Status | Three parts: API, Multimodal Live API and Google AI Studio |
| Vertex AI on Google Cloud | Google Cloud Service Health | A product named Vertex Gemini API, plus Gemini Enterprise and Gemini Code Assist |
We read all three on 15 November 2026. The Workspace dashboard showed "No incidents", updated 06:35 UTC, and the AI Studio page showed "All Systems Operational". An evergreen page can't tell you today's status, so open them yourself. It takes under a minute.
Watch for lookalikes. Gemini is also a crypto exchange, and several of the live-status sites that rank for this query track the exchange, not Google. Google's dashboard also says "If you are experiencing an issue not listed here, please contact Support", so a missing entry doesn't clear every problem.
What do the status labels mean?
The Workspace dashboard uses four labels: Available, Service information, Service disruption and Service outage. Neither the dashboard nor its help page defines them, so we read what Google put under each one.
The labels don't track how long an incident lasted. Google put the May 2026 incident, with "Something went wrong(13)" errors in conversations, under Service information, and it ran about 98 hours. The 10 June 2026 incident ran 7 hours on the dashboard and carried Service disruption. The February one, with chat history hidden from the side navigation, was Service disruption and ran about 100 hours. None of Gemini's four incidents in that history carried Service outage.
So read the incident text, not the icon. It always says more. Each entry gives a symptom, a cause once Google knows it, and a workaround. For both the May and June incidents, the workaround line read "None at this time." Google also says "A final update is posted when the issue is resolved", so an entry with no final update is still open.
The AI Studio page works differently. It shows an overall banner, three parts with a 90 day bar each, and a list of past incidents. Each incident moves through steps Google words as Detected, Identified, Update, Mitigated and Resolved. "Identified" means the cause is known and a fix is on its way: on 1 October 2026 Google wrote "We have identified the root cause and a fix is being rolled out."
That page also has a filter for tier. Google's own line under it reads "Free-tier requests use sheddable capacity, while billed-tier requests are protected by critical priority." In plain terms, when capacity is short, free API traffic is the part Google can drop first. A free key can fail while a billed one keeps working.
How often do Gemini outages happen, and how long do they last?
The Workspace dashboard lists four Gemini app incidents among its 32 from 14 November 2025 to 25 September 2026, about one in eight. Two ended within 8 hours (7.2 and 7.4). Two ran about four days (98.2 and 99.6), for roughly 212 hours in total. Here is each one, plus the latest on the other two pages, with Google's own dates.
| Dates | Page | What users saw | Google's label or step | How long |
|---|---|---|---|---|
| 18 to 22 February 2026 | Workspace dashboard | Chat history hidden from the side navigation | Service disruption | About 100 hours |
| 27 February 2026 | Google Cloud | More errors on the Vertex AI Gemini API global endpoint | Service information | 118 minutes |
| 4 to 8 May 2026 | Workspace dashboard | "Something went wrong(13)" in conversations | Service information | About 98 hours |
| 10 June 2026 | Workspace dashboard | "Something Went Wrong" errors on web, mobile and Chrome | Service disruption | 7 hours on the dashboard; 14 hours 49 minutes to full restoration, per Google's report |
| 11 September 2026 | Workspace dashboard | Intermittent errors on Gemini 3.1 Pro | Service information | 7 hours 22 minutes |
| 14 to 16 September 2026 | AI Studio | Batch requests not finishing in 24 hours | Detected, then Resolved | About 45 hours between the two entries |
| 30 October 2026 | AI Studio | All requests using API keys failing | Detected, then Resolved | 2 hours 36 minutes between the two entries |
The 30 October entry is the clearest case of a full API outage. Google wrote "All requests using API keys are currently failing." It stayed open for 2 hours 36 minutes by the page's own timestamps. During that time the Gemini app on the Workspace dashboard was not in any listed incident, because the API and the app are tracked on separate pages.
Google's June report on the 10 June incident gives the sharpest number. It says the elevated error rate lasted about 6 hours 55 minutes, while "full service restoration" took 14 hours 49 minutes in total. If the icon turns green and your chat still fails, that gap is usually why. Give it an hour.
Is Gemini not working right now, or only for you?
Gemini is only down for you when the status pages are green and the error stays on one device, one account or one chat. Run these three checks, which take two minutes and split the cases cleanly.
| Test | If it works | If it fails too |
|---|---|---|
| Open gemini.google.com on your phone with wifi off | Your network, VPN or router | Google, your account, or the chat |
| Open a new chat and send "hi" | The old chat is stuck | The account or Google |
| Sign in with a second Google account | Your first account is blocked | Google or your network |
The wording of the error helps. At sign in, Google's help page says "If you get a 'Something went wrong' message while trying to access or sign in to the Gemini web app, your account can't access the app at the moment." Google lists "location, age, or type of account" as causes, and says to try again later. That one is about you, not an outage. Don't wait it out.
Numbered errors in the middle of a chat are different. Google tied the (13) code to its May 2026 incident. If other people report the same number in the same hour, treat it as Google's. Our guide to Gemini's something went wrong error lists each code and where it was first reported.
A message about limits is also not an outage. If Gemini says you've run out of prompts, the fix is a reset or a bigger plan, and Gemini usage limit reached explains each one.
For API calls, the status code tells you which side to blame. Google's API errors page lists 429 quota_exceeded as "You have exceeded your daily quota", which is your account. It lists 503 service_unavailable as "The service is temporarily overloaded or down", which is Google's.
Why is Gemini not working when every status page is green?
A green page means Google hasn't posted an incident. It doesn't mean your request will go through. Google's Chrome help page says it plainly for one case: "Turn off your VPN." Work down this list and stop at the first step that helps.
- Copy your prompt out of the box. A reload can wipe it, and you'll want it for step 4.
- Send it again once. During the 10 June 2026 outage, Google's report says users saw a 50% error rate on prompts, so half still went through.
- Open a new chat and send a short text prompt. No file, no link. If that works, the old chat is the problem.
- If it still fails, use another assistant for now and come back after the next status check.
Device, browser, VPN and account fixes are not repeated here. Gemini not working walks through them, and Gemini something went wrong matches each error code to its fix.
For API code, don't retry by hand. Google's troubleshooting guide says to retry only transient errors such as 429, 408 or 5xx, with exponential backoff. It says the Python SDK retries up to four times, starting at about 1 second and capping at 60 seconds. Don't retry 400, 402 or 403, because those mean a bad key, balance or request.
If this keeps happening, subscribe instead of checking. Google's help page says the RSS feed gives "the fastest outage alerts", and the dashboard also offers a JSON history for scripts. Admins can add an Apps outage alert rule in the Admin console.
The wider picture is in AI chatbot outages, which counts incidents across ChatGPT, Claude, Gemini and others. Gemini not working covers the device side: Android Auto, watches and Chrome.
What can you use while Gemini is down?
Another vendor, and a copy of your prompt. Don't sit and refresh. An incident on Google's side won't shorten for any workaround you try, so move the task. If the work is in a chat, paste a short summary of the thread into Claude or GPT and keep going. What to use when ChatGPT is down and Is Claude down cover the other two.
Whizi, which publishes this page, puts Claude, GPT and Gemini in one picker on Pro, so the switch is a menu click and the thread comes with you (how switching works). The first week of Pro is $1.99. Where it loses: Whizi does not fix a Google outage. A Gemini model in the picker is still Google's, so during an incident the move is to a Claude or GPT row. The free plan gives 10 messages a day on Whizi AI, Auto and Claude Haiku 5.5, with no card, but no Gemini model.
Frequently asked questions
Is Gemini down right now?
Check the Gemini row on Google's Workspace Status Dashboard for the app, and aistudio.google.com/status for the API. When we read them on 15 November 2026, the dashboard showed "No incidents" and AI Studio showed "All Systems Operational". If both are green, run the two minute tests above before blaming Google. Gemini not working covers the device fixes.
Is status.gemini.com Google's status page?
No. Gemini is also a crypto exchange, and several live-status sites that rank for this query track the exchange. Google's Gemini is tracked on the Workspace Status Dashboard, aistudio.google.com/status and status.cloud.google.com. AI chatbot outages compares how each vendor posts incidents.
Why does Gemini say something went wrong?
At sign in, Google says your account can't access the app right now, for reasons that include location, age or account type. Mid chat, it can be an outage: Google tied "Something went wrong(13)" to a May 2026 incident. Gemini something went wrong matches each code to its fix.
Does paying stop the Gemini API from failing?
Not entirely, but the status page says billed traffic is protected first. Its text reads that free-tier requests use sheddable capacity, while billed-tier requests are protected by critical priority. On 30 October 2026 the page logged an outage affecting all traffic using API keys. Gemini API rate limits explains the quota side.
What is going on with Gemini today?
Only the Gemini row on Google's Workspace Status Dashboard and aistudio.google.com/status can say, and this page can't. For context, the dashboard's JSON history lists four Gemini incidents from 14 November 2025 to 25 September 2026, the latest on 11 September 2026 with intermittent errors on Gemini 3.1 Pro. Gemini not working covers fixes on your side.
Is there a global outage right now?
Google's pages cover only Google, so a wider internet outage shows up there only as a side effect. On 18 November 2025, for example, aistudio.google.com/status logged apps in Build that did not load because of a global Cloudflare outage. AI chatbot outages lists where each vendor posts incidents.