Google has a growing Gemini problem that readers find hard to ignore

We’ve seen growing reports of users suggesting that Google Gemini, including its models and its virtual assistant face, is getting “dumber.” Various threads on Reddit highlighted this problem, with some noting frequent misunderstandings, degraded performance, and virtually no perceptible performance differences between the various premium tiers.

To gain a better understanding of whether Google Gemini really is facing a performance decline, we asked our readers for comment in a recent poll. After more than 1,200 votes received, the verdict is in.

Over 60% of respondents report that Google Gemini has gotten worse in recent months, at least according to their experience. When disregarding the 5.6% of users who don’t use Gemini at all, disgruntled Gemini users account for just under 65% of the vote, or nearly two in every three respondents.

Not everyone believes that the Gemini experience is regressing, with just under 20% voting “No, it hasn’t,” in our poll. A further 14.6% aren’t sure, which suggests that there’s no perceptible improvement or decline either way.

‘Gemini is just frustrating’

Our readers also left their impressions of Gemini in the comments, and it don’t make for positive reading.

Reader aguygton1215 explains their move to a competing AI service:

Dec 2025 was the last time I thought Gemini was actually good and competitive. I use Claude Code extensively for work every day, that has its own complaints but compared to Gemini Pro today, it’s miles ahead. I[‘d] rather ask my work Claude account on my Pixel 10 Pro XL to create tasks or look up places than use the Gemini Pro I got for a year. Gemini is just frustrating, and barely better than Google search results. Why even use it when it’s more effort than Google search or just doing the task myself?

Commenter Kevin Fleming, who praised Gemini for the “value it provided,” highlights its recent drop in performance and response quality:

However, not only has it gotten dumber, or stupider, the responses are clearly configured to use less compute and thus I get much less comprehensive responses, by a factor of about five would be my estimate. The value is now reduced by that amount. It’s almost not worth my time.

Since we’re all different with various expectations, use cases, and requirements, it’s likely that Gemini will work adequately for some and fall well short for others. It’s also deployed in various ways across Google’s many service offerings. However, it is worth evaluating the efficacy of that tool when a large portion of users are raising concerns.

Have you noticed any issues with Gemini’s understanding, speed, or output in recent months? What do you use it for, and how do you use it? Let’s continue this discussion in the comments section.