Users complain about Gemini as Google promises better reliability

More users of Google's AI assistant Gemini say it has gotten worse. That is according to user threads in Google's developer forum and coverage of the complaints. These are users' own accounts, not independent measurements, and it is unclear how widespread the problems are.

What users describe

Users cite long response times, conversations that stall mid-answer, and Gemini making up incorrect information, known as hallucination. Some also say Gemini loses the thread in long conversations.

In a thread in the developer forum, one user wrote on April 19, 2026, that Gemini rushes, skips steps, edits the wrong file versions and forgets earlier context. The user felt Gemini performed worse than its competitor Claude. Two other users confirmed similar problems in early May.

Other threads mention a drop in performance as far back as January 2026 and a decline in coding quality in July.

New limits sparked anger

Around May 17, 2026, Google introduced new usage limits for Gemini, including throttling of the Flash model. Users criticized the fact that it happened without advance notice, and some demanded refunds.

According to Workalizer, Google employee Vijay Sivasubramanian confirmed that users are experiencing slow responses, conversations that stall, more hallucination and usage limits being reached too early. He also said it is not possible to go back to older Gemini versions, and that model updates are managed globally by Google. He asked users to send detailed feedback in the app.

Google asks what is falling short

Josh Woodward, VP of Google Labs, asked users on X to write what Gemini still does poorly. It drew more than 1,700 replies, according to WebProNews.

The most frequently cited problems were unreliable integration with Workspace, including Gmail and Docs (109 votes), errors when Gemini needs to use external tools, and poor organization of chats. Others complained about loss of context, calculation errors, hallucinations, dictation problems and bugs in the mobile app.

According to the coverage, Woodward acknowledged that reliability needs to improve and that improvements are already under way. Google says noticeable improvements to tool calling are coming soon and that chat organization will be rethought. These are plans, not something that has happened yet.

Has Gemini been made worse on purpose?

A theory is circulating online that Google has deliberately made Gemini worse. No confirmation from Google has been found, and the theory is speculation. New usage limits and server load are possible explanations for the slowness, but there is no documentation that they are the cause.

For everyday users, this means Gemini may feel less stable for a while, especially in long conversations and when it has to work with Gmail and Docs.