A November 2026 reader poll drawing more than 1,200 votes suggests many users believe Google Gemini and the wider Gemini model family have declined, with a little over 60% of respondents saying Gemini had gotten worse—a figure that climbs to just under 65% when non-users are excluded. Respondents cited more misunderstandings, weaker answers, a worse assistant experience overall, and paid tiers that don’t seem to offer much, while just under 20% disagreed and 14.6% said they weren’t sure. Reliability issues likely contributed: August brought latency, endless loading screens, and Error 1076 reports; October saw a content-generation outage; in February, a Vertex issue caused API errors for nearly two hours; and on October 9, Google introduced automatic thinking-level routing, which can make speed and response quality feel uneven from one prompt to the next. Google has kept updating Gemini, launching Gemini 4 Carbon on October 28, 2026, along with API patches, a reported score of 59 on the Artificial Analysis Intelligence Index, according to Artificial Analysis, and lower per-task costs than some rivals. Still, independent analysis pointed to a gap between benchmark scores and everyday use, with reported AA-Omniscience hallucination rates of 88% for Gemini 2.5 Pro and 91% for Gemini 3 Flash; Gemini also ranked last in one source-citation comparison.