Is Gemini down right now?
This page shows the current status of Google Gemini, available at gemini.google.com and through the Google AI Studio and Vertex AI APIs. The indicator at the top is refreshed every 60 seconds. Green means the consumer chat product is responding normally. Amber means it is degraded — typically slow generations, missing features, or partial regional issues. Red means a major incident is in progress and a meaningful number of users cannot use the product.
Gemini is unusual among major AI assistants because Google operates it across a much wider set of surfaces than its competitors — gemini.google.com, the Gemini integration in Workspace (Gmail, Docs, Meet), the Pixel device assistant, and the AI Studio and Vertex APIs — and any of these can be down independently. The status indicator on this page tracks the consumer-facing Gemini product, which is the one most users mean when they say "Gemini is down."
Gemini outage history and common causes
Gemini's failure modes are different from its competitors because it is hosted on Google's own cloud infrastructure rather than a third-party cloud. That means Google Cloud incidents propagate to Gemini almost immediately. The biggest Gemini outages of the past year have all coincided with broader Google Cloud or Google services incidents.
Specific recurring causes include: capacity issues during major model launches (the move from Gemini 1.5 to 2.0 and 2.5 caused short bursts of unavailability), authentication problems that lock users out of gemini.google.com when the underlying API is fine, regional outages tied to specific Google data centers, and feature-level breakage where the chat works but specific tools (Deep Research, image generation, code execution) fail. Workspace integrations have their own failure surface and can break even when the standalone Gemini product is fine.
One pattern worth knowing about: because Gemini is gated by Google account access, a problem with Google Sign-In can manifest as "Gemini is broken" for an enormous number of users at once even though the model serving infrastructure is healthy.
Is Gemini reliable?
Gemini benefits from running on Google's mature cloud infrastructure, which has uptime characteristics roughly comparable to Gmail or Google Search at the platform layer. The actual model serving sits on top of that, which adds its own failure surface. Independent monitoring tracks 30-day availability for the consumer Gemini product in the 99.5 to 99.9 percent range — broadly competitive with ChatGPT and Claude, though not dramatically better despite the underlying infrastructure advantage.
For enterprise use, Vertex AI — which serves Gemini through Google Cloud — has stronger uptime guarantees and more transparent SLAs than the consumer chat product. If you are integrating Gemini into a real workflow, the Vertex API is the path with the better operational story.
What to do when Gemini is down
First, check whether other Google products are also misbehaving. If Gmail, Docs, or Search are also affected, the issue is at Google Cloud, not specifically Gemini, and there is nothing user-side to do but wait. If only Gemini is down, the issue is contained to the AI product.
Second, try a different surface. If gemini.google.com is broken, AI Studio (aistudio.google.com) often still works because it is a separate frontend. The mobile Gemini app is on a separate release pipeline. The Workspace Gemini integration usually works when standalone Gemini is broken, and vice versa.
Third, switch tools. ChatGPT and Claude can handle most of the same tasks. Perplexity is a strong substitute for the search-augmented question answering that Gemini is good at. For coding work specifically, GitHub Copilot runs on different infrastructure than the consumer chat assistants.
Finally, report the outage on this page. User reports get aggregated into the counter above and help the next person who lands here figure out whether the problem is real or just on their end.