Google AI Mode now uses Gemini 3.8 Flash, but it does not seem as link or citation friendly, according to Search Engine Roundtable's October 28 recap. The observation matters because the model powering AI Mode shapes how often answers cite sources, and how prominently those citations appear.
AI Mode is Google's conversational search experience, distinct from AI Overviews, and its underlying model determines the style, depth, and sourcing of its answers. The move to Gemini 3.8 Flash brings a faster, lighter model to the surface, which suits conversational latency but may trade away some of the citation richness of a larger model.
The recap's wording is cautious: AI Mode does not seem as link or citation friendly on the new model. That is an observed impression from usage, not a measured study, and publishers should treat it as a hypothesis to test rather than a confirmed regression.
Different models have different habits: how many sources they consult, how many they cite, whether they prefer inline links or a sources panel, and how much they summarise versus quote. A model optimised for speed and fluency may produce smoother answers with fewer visible citations, even if its underlying retrieval is similar.
That variability is why citation tracking must be model-aware. An AI Mode answer in one month and an AI Mode answer the next month may come from different models, and comparing their citation counts without noting the model change will mislead.
Run a fixed set of your money queries through AI Mode regularly, and record which sources are cited, how many citations appear, and whether your pages are among them. Keep the query set stable so changes in citation behaviour show up against a constant baseline.
If your citation rate falls after a model change while your content is unchanged, the likely cause is the model, not your pages. That distinction saves wasted rewriting: the fix for a model-driven citation drop is patience and continued authority building, not a hasty content overhaul.
Google's October 15 announcement promised more prominent links in AI Mode and AI Overviews, with links placed higher, thumbnails on related links, and tests highlighting official sites. A model change that reduces citation friendliness would work against that promise if it rolled out widely.
The two developments together suggest Google is tuning AI Mode on multiple dimensions at once: link presentation, model choice, and ad blending. Publishers should expect citation behaviour to keep shifting while these tunings interact, and should avoid overreacting to any single week's citation data.
Model churn is a reason to build citation-worthy content, not to chase a single model's quirks. Pages with original data, clear answers, and strong external validation are the pages most models cite, across model generations.
Track citations by surface and by date, note model changes when they are reported, and judge your content on a multi-week trend rather than a single day's answers. That discipline turns model noise into a signal you can actually act on.
Every model choice trades speed against depth. A lighter, faster model answers conversational queries with less latency, which matters when users expect instant replies, but it may consult fewer sources and cite less thoroughly than a heavier model given the same question.
Google's challenge is to keep AI Mode feeling instant while keeping its answers well-sourced, and the Flash move suggests speed is winning that trade-off for now. Publishers should expect citation density in AI Mode to vary with model choices, and should judge their citation performance over months rather than days. Note the model version alongside each measurement so later comparisons stay honest over time.
Sources: Search Engine Roundtable podcast recap, October 28, 2026.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial