Google Flips Core Search to Gemini 3.5 Flash, Making AI Mode the Default

Google has finished rolling out Gemini 3.5 Flash as the default model behind Google Search's AI Mode, completing a transition that effectively retires the traditional ten-blue-links search box for the majority of users. AI Mode is now the default experience for global Search, and Gemini 3.5 Flash is what runs under it — for every user, free or paid.

The switch had been previewed at Google I/O 2026 in mid-May and went global over the following two weeks. As of this week, when most users type a query into Google, the response that comes back is a conversational, multimodal AI answer, not a list of links. Links still appear, but they sit below the AI response, not above it.

What's actually different

AI Mode is a fundamentally different interaction than traditional Search. You can ask longer, more detailed questions. You can upload screenshots, PDFs, photos, and videos directly into the search box. You can follow up with contextual prompts that reference the previous answer. Search now behaves like a chat assistant that happens to know the web.

The switch to Gemini 3.5 Flash makes that experience fast enough to scale to billions of queries. Google says 3.5 Flash outputs tokens four times faster than other frontier models at its tier, while delivering what Google describes as "Pro-level" reasoning. That's the trick: get most of the quality of a Pro model at Flash-class latency and cost, so it can run on every query without breaking the unit economics of Search.

Gemini 3.5 Flash also powers the Gemini app's free tier — meaning the same model now sits under both Google's search index and its standalone chatbot. That consolidation simplifies Google's stack and makes the experience more consistent across surfaces.

Why this matters

Search has been Google's most defended business for two decades. Flipping the default interface is the biggest change to that product since Universal Search in 2007. The implication for the open web is enormous: when AI Mode answers the question inline, users click through to source pages far less often. Publishers, e-commerce sites, and content sites built around Google referral traffic are going to feel this in the next two quarters.

For users, the practical effect is a Search that handles complex, multi-step tasks the old keyword box couldn't. Comparing two products with similar specs, summarizing a long PDF you just uploaded, asking a follow-up that depends on the previous answer — all of these now work natively in the search box.

For Google's competitors, the move forces a strategic choice. ChatGPT and Claude have built their consumer products around the chatbot interface; Google has now ported that interface into the default Search experience. The differentiation question for the AI assistant category just got harder.

The model behind the curtain

Gemini 3.5 Flash sits in the middle of Google's current model lineup. Above it is Gemini 3.5 Pro and the still-limited Gemini Omni — Google's multimodal flagship that handles audio, video, and real-time interaction. Below it is the lightweight Nano family for on-device inference.

What makes 3.5 Flash fit Search is the latency-quality combination. Search queries need to feel instant. Earlier attempts at AI-powered Search (Google's own SGE preview, Microsoft's Bing Chat) struggled because reasoning-capable models were too slow to feel like Search. 3.5 Flash is the first model Google has shipped that can sustain Search-grade response times across the global query load.

Google also rolled out Gemini Spark — a persistent 24/7 personal agent — in beta to Ultra-tier subscribers alongside the Search switch. Spark is the more ambitious bet: an agent that runs continuously in the background, handling delegated tasks across email, calendar, and the web. The AI Mode rollout is what most users will notice this month; Spark is what Google is testing for the year after.

Publisher and SEO industry reaction

Reaction from the publishing industry has been alarmed. AI Mode answers questions inline using snippets, summaries, and synthesis from web sources — but it dramatically reduces the click-through rate to those sources. Major publishers have spent the last year preparing for this transition, but the actual switchover has still landed hard.

The SEO industry is in restructuring mode. Optimizing for AI Mode requires different signals than ranking in the old ten-blue-links Search. Structured data, schema, source authority, and machine-readable summaries matter more; keyword density and link-building matter less. The shift from "ranking" to "being cited" is the new mental model the field is converging on.

E-commerce sites are seeing the cleanest impact. Product comparison, spec lookup, and shopping intent queries are exactly the kind of multi-step task AI Mode handles best, and exactly the queries that used to drive paid search clicks.

What to watch

The first thing to watch is what happens to Google's ad business. Search ads are still the company's largest revenue stream. AI Mode changes where ads appear and how often they get clicked. Google has said sponsored placements will continue to surface in AI Mode, but the unit economics of an AI-summarized SERP are very different from a traditional results page.

The second thing to watch is publisher pushback. There's already pressure on Google to give publishers more granular control over whether their content can be used in AI Mode summaries — and a regulatory environment, especially in the EU, that's increasingly receptive to that pressure.

Bottom line

The Search box you grew up with is gone. What replaces it is faster and smarter for the things you'd actually use a search engine for. But the deal it offers the open web — clicks in exchange for being indexed — has just been rewritten. For publishers, that's an existential adjustment. For users, it's an upgrade you'll feel within the first three queries. And for Google, it's the biggest product change since Search itself was invented.