Google IO 2026: Gemini Spark, New Models, and a 50 Dollar Price Cut

📖 6 min read

Google wrapped its annual I/O developer conference on May 20, 2026, and the headline isn’t one feature – it’s an entire ecosystem push. In roughly 2 hours, the company announced new AI models, a revamped app, an always-on background agent, AI-powered video generation, a price cut on its most expensive plan, and smart glasses coming this fall. Here’s what actually matters and what to do about it.

The Numbers First

• 900 million people now use Google’s Gemini assistant
• 50 billion images generated with Gemini to date
• 9.7 trillion tokens processed per month on the platform
• AI Ultra plan price cut: from $250/month down to $200/month
• New mid-tier AI Ultra plan: $100/month

What Was Actually Announced

Gemini 3.5 and Gemini 3.5 Flash

Google released two new models at I/O: Gemini 3.5 Pro and a faster, cheaper version called Gemini 3.5 Flash. Both are available immediately in Google Search and the Gemini app. According to Google, Gemini 3.5 Flash outperforms its predecessor Gemini 3.1 Pro on coding and agentic benchmarks, while being significantly more cost-efficient. More models in the Gemini 3.5 family are expected next month.

Gemini Spark – The Always-On Agent

This is the announcement most people should pay attention to. Gemini Spark is a background AI agent that runs continuously while you go about your day. It can send emails on your behalf, scan your monthly credit card statements for hidden subscription fees, summarize meeting notes, and connect with Workspace apps (Docs, Sheets, Slides) as well as third-party apps like Canva, Instacart, and OpenTable. It runs on Gemini 3.5 Flash.

The concept is “agentic AI that acts without being asked” – closer to an automated employee than a chatbot. Whether that feels helpful or invasive will depend on the person.

Daily Brief

Google is adding a personalized morning digest that pulls from your calendar, Gmail, and connected apps to summarize your day. It prioritizes based on your goals and accepts thumbs-up/thumbs-down feedback to improve over time. Google had tested something similar in Google Labs in December 2025, but this is the full rollout.

Gemini Omni – AI Video Generation

Google’s answer to OpenAI’s now-shuttered Sora 2 is called Gemini Omni. It accepts any combination of images, audio, video, and text as input and generates realistic video output. The demo showed a user walking through a metal sculpture on video, then asking the AI to transform the sculpture into bubbles in real time. Google says physics simulation is significantly improved over previous generations. There is also an Avatars feature that creates videos featuring a digital version of yourself using your own voice.

Search Overhaul – Agents in the Box

Starting today, Google is rolling out what it calls the “intelligent search box” to all users. Ask it a complex question and it now generates contextual answers, images, or short video clips inline. A Generative UI feature creates dynamic layouts based on the content type returned. This summer, the full rollout of these agents goes to everyone.

Ask YouTube and Docs Live

Ask YouTube lets you pose natural language questions (“how do I fix a leaking tap?”) and an AI agent finds relevant YouTube videos, including jumping to the exact timestamps that answer your question. Docs Live lets you dictate your intent with your voice and an AI agent writes, edits, and pulls in citations from the web. Both feel useful on paper – execution will determine whether they’re actually faster than existing workflows.

Android XR Smart Glasses

Google confirmed Android XR-powered smart glasses are coming in fall 2026. No pricing was announced. They were briefly demoed but not the focus of the keynote.

Pricing Breakdown

Plan	Old Price	New Price	Notes
Gemini AI Ultra (top tier)	$250/month	$200/month	$50/month savings
New AI Ultra (mid tier)	N/A	$100/month	New plan – access to Gemini 3.5 models
AI Pro	Available	Unchanged	Standard Gemini access

The $50/month cut on AI Ultra is notable. At $200/month you’re getting access to Gemini Spark, Gemini Omni, and priority access to the 3.5 models. For power users or small business owners who live in Google Workspace, the math gets more interesting.

What This Means in Practice

Google’s strategic direction is clear: it wants Gemini to be the operating system layer for your productivity, not just a chatbot you open occasionally. The shift from “ask a question, get an answer” to “an agent that works in the background without you asking” is a meaningful architectural change.

The risk is real though. Users trusting an AI to send emails, manage subscriptions, and act autonomously is a high-stakes bet on accuracy. One misread instruction or one rogue email sent to the wrong person creates exactly the kind of backlash that slows adoption. Sundar Pichai acknowledged this himself, noting that this is the point in the cycle “where people want to see real value in the products they use on a day-to-day basis.” That’s a careful way of saying Google knows the demos need to hold up in the real world.

Google is also betting that being embedded in Search gives it an advantage over standalone AI tools like ChatGPT. With 900 million Gemini users already, the distribution is there. The question is whether the product quality matches the scale.

What Competitors Are Doing

Company	Background Agent	Video Generation	Top Plan Price
Google (Gemini)	Gemini Spark (new)	Gemini Omni (new)	$200/month
OpenAI (ChatGPT)	Operator mode	Sora 2 (shut down)	$200/month (Pro)
Anthropic (Claude)	Limited agent tools	None	$100/month (Max)
Meta (Llama)	No	No	Free (open source)

What To Do About It

• If you pay for Gemini AI Ultra at $250/month: your bill drops to $200 automatically. No action needed.
• If you’re on the fence about upgrading: the new $100/month AI Ultra tier is likely the sweet spot. You get Gemini 3.5 access and Gemini Spark without committing to the top tier.
• If you use Google Workspace heavily (Docs, Gmail, Calendar): Daily Brief and Docs Live are worth testing this summer when they roll out broadly. They address real friction points in productivity workflows.
• If you use Gemini only casually: The free tier is getting meaningfully better with Gemini 3.5 Flash showing up in standard Search. You don’t need to pay for anything new right now.
• If you’re a developer: Gemini 3.5 Flash’s improved performance on agentic benchmarks makes it worth re-evaluating against GPT-4.1 and Claude Sonnet for API integrations.

What Google Didn’t Say

Notably absent: specific benchmark numbers comparing Gemini 3.5 to competing models. Google said the model “outperforms” and “is faster” without giving standardized scores (MMLU, HumanEval, etc.) that would allow direct comparison. Also quiet: any clarity on how Gemini Spark handles edge cases, errors, or user consent flows when acting autonomously. That will matter a lot.

BetOnAI Verdict

Google I/O 2026 is the most substantive AI product event Google has run in years. Gemini Spark and the Generative Search overhaul are not incremental updates – they represent a real attempt to shift how people work with AI from reactive to proactive. The price cuts on AI Ultra make the higher tiers more defensible to consider.

That said: Google has a consistent track record of announcing ambitious AI features that take longer than expected to reach the quality shown onstage. Gemini Spark’s value entirely depends on how well it handles real-world ambiguity. We’d rate this announcement an 8/10 on ambition, with the real grade to be determined when the summer rollout lands.

For now, if you live in Google’s ecosystem, the $100/month AI Ultra plan is worth a trial run when it becomes available. If you’re outside Google’s ecosystem, there’s nothing here compelling enough to switch.

---

Sources:

• Wired – Everything Announced at Google I/O 2026
• NewsBytesApp – 5 big Gemini upgrades Google just announced
• Tom’s Guide – Biggest Google I/O 2026 announcements
• 9to5Google – Everything Google announced at I/O 2026