Google I/O General Assembly All in AI: Gemini Model, the largest adaptation of the search box in 25 years, the return of the Ai glasses, the Spark smart body premiere

Original by: Li Dan

Original source:See you on Wall Street

On Tuesday, 19 Eastern American time, Google, under the Alphabet flag, issued an intensive AI update on the Annual Developer Conference Google I/O 2026, from redesigning Google searches to publishing Gemini Omni and 3.5 Flash models and Gemini Spark smarts to opening a new generation of AI glasses for the first time. Google is trying to embed AI into the entire Internet access system — search, Android, office, video and even the next generation of hardware。

Alphabet and Google CEO Sundar Pichai revealed in their keynote address that the number of monthly live users of the Gemini application (App) has doubled to over 900 million a year; that the “AI Overview” of Google Search has now exceeded 2.5 billion active users in the month; and that the “AI mode” has reached more than 1 billion users。

Pichai stressed that AI was no longer just a chat robot, but a “new generation computing platform”. He said to the General Assembly: “AI is lighting every part of the company.”

The Gemini Omni model, Gemini 3.5 Flash, to the Gemini Spark smart body able to perform its tasks on a sustainable basis, and to the realignment of the AI subscription system at this Congress show that Google is trying to upgrade Gemini from a chat robot to an “AI operating system” covering search, office, development, video and mobile ecology。

The Conference was also seen by the outside world as a comprehensive response by Google to competitors like OpenAI and Microsoft. Over the past year, as ChatGPT has changed user search and content consumption habits, Google's traditional search operations have faced the greatest challenge since their inception, while the I/O Congress has clearly demonstrated its latest strategy of using the world's largest Internet traffic portal to embed AI's “pretochemical” into the entire Google ecology。

SEARCH FOR THE LARGEST ADAPTATION IN 25 YEARS: THE SEARCH BOX BECOMES THE AI PORTAL

ONE OF THE CORE CHANGES OF THE CURRENT I/O IS THE COMPLETE AI-BASED ADAPTATION OF GOOGLE TO SEARCH OPERATIONS。

GOOGLE IS REPORTEDLY REDESIGNING ITS LANDMARK SEARCH BOX TO ENABLE IT TO BETTER DEAL WITH “LONGER AND MORE COMPLEX ISSUES” THAT USERS ASK THE AI CHAT ROBOT。

The new search box will:

• Support more natural language input
• Improved access to images and documents
• Auxiliary user improvement search
• SUPPORTING AI ' S INTELLIGENT BODY IN ITS ONGOING MISSION。

Nick Fox, Head of Google Knowledge and Information Platform, said: “Users should be able to enter any problem in their minds directly into the search box.”

Google also plans to add the Agent feature to the search to help users: follow up on topics; book services; manage health information; and create personalized task panels. Some of these functions will be open to subscriptions on a fee basis only。

ACCORDING TO INDUSTRY, THIS MEANS THAT GOOGLE SEARCHES ARE MOVING FROM THE TRADITIONAL “LINK DISTRIBUTOR” TO THE AI MISSION PORTAL。

Google's strongest multi-modular model, Gemini Omni

One of the central updates of the Congress was Google ' s launch of the new Gemini Omni model。

Several scientific and technical media reports indicate that Gemini Omni has been identified as the most advanced multimodular AI system in Google, capable of processing text, images, videos, voice and real-time interactive missions simultaneously。

The biggest change in Omni compared to the previous Gemini series was:

• More emphasis on real-time interaction
• Better suited to the Agent scenario
• Continuous missions can be carried out across applications and equipment。

It is believed that Google is using this to directly target the GPT-5 Agent system at OpenAI and the copilot ecology being advanced by Microsoft。

IN THE GENERAL ASSEMBLY, GOOGLE EXECUTIVES HAVE REPEATEDLY STRESSED THAT “AI DOES NOT JUST ANSWER QUESTIONS, BUT BEGINS TO REALLY FINISH THE JOB”. THIS CHANGE MEANS THAT GOOGLE'S AI STRATEGY HAS SHIFTED FROM A “CHAT ASSISTANT” TO A “MISSION IMPLEMENTATION PLATFORM”。

Gemini 3.5 Flash primary hit speed and low cost & nbsp; 3.5 Pro released June

In addition to the flagship model, Google launched Gemini 3.5 Flash。

According to I/O General Assembly information and Google Developer's page, the Flash series is located at low delay, high response speed and lower reasoning costs, focusing on developers; mobile ends; real-time interaction; and AI programming and automation scenarios。

As a lightweight new member in the Gemini model, Gemini 3.5 Flash has cutting-edge technical capabilities and pricing is only half, and in some cases as low as one third, of the same front model. Google announced that Gemini 3.5 Flash would be a global default model of Gemini App and the "AI mode" in Google search。

Google claims that Gemini 3.5 Flash's cyber-security protection has been enhanced, making it “lower likely that harmful content will be generated and less likely that security requests will be denied”。

Google also revealed that its more powerful version, Gemini 3.5 Pro, is currently available for internal use, but will not be released to a wider audience until next June。

ACCORDING TO THE MEDIA, GOOGLE IS CLEARLY AWARE THAT THE AI COMPETITION IS NOT JUST “WHO IS THE SMARTEST”, BUT WHO IS FASTER, CHEAPER AND MORE SUITABLE FOR LARGE-SCALE DEPLOYMENT。

The importance of delay and cost is rapidly increasing in the Age of Age and Real Time AI。

Gemini 3.5 Flash is reported to be one of the important foundation models for the new Google Generation AI subscription system。

Gemini Spark: Google official bet "AI Smart"

The market is more concerned with Gemini Spark than with model updates。

GOOGLE DEFINES IT AS A “CONTINUOUS AI INTELLIGENT BODY” CAPABLE OF CARRYING OUT TASKS IN THE BACKGROUND FOR A LONG PERIOD OF TIME. ACCORDING TO SEVERAL MEDIA OUTLETS, THIS IS ONE OF THE MOST STRATEGIC PUBLICATIONS OF THIS I/O SESSION。

According to Google, Gemini Spark will be able to:

• (b) Automatically organize mail
• Monitoring of user schedules
• Tracking web-based information
• Manage documents
• Completion of cross-application workflow
• Even partially consumed and booked on behalf of users。

According to the report, Google is promoting the concept of “Always-on-line” that AI no longer waits for instructions from users but takes the initiative to handle tasks。

According to Josh Woodward, Vice-President of Google, “Spark represents a major shift for Gemini.”   industry believes that this means that the AI industry is moving from the “chat robotic age” to the “smart age”。

Previously, ChatGPT and Gemini were essentially “dialogue tools”; and Spark was closer to real digital assistants。

Google clearly wants to turn Gemini into an AI implementation portal at the Internet level through Gmail, Chrome, YouTube, Docs and Android ecology。

"We're entering the age of search intelligence."

"We're entering the age of searching for intelligence." Google searches have been able to leapfrog performance with the Gemini 3.5 model, enabling them to perform complex tasks such as making travel itineraries。

The Vice President Robbie Stein of Google Search demonstrated Gemini's intelligence agency capabilities. He said, “We are convinced that the best search experience should be one that really works for you.”

THROUGH A NEW GOOGLE SEARCH FOR “AI OVERVIEW”, USERS WILL GAIN VERY INTERACTIVE EXPERIENCES DURING THE SEARCH — FOR EXAMPLE, BY EXPLORING IN DEPTH THE MYSTERY OF A BLACK HOLE. FOR DIFFERENT SEARCH TOPICS, THE SEARCH FUNCTION CAN EVEN AUTOMATICALLY GENERATE ANIMATED PRESENTATIONS AND CUSTOMIZED USER INTERFACES。

Google searches have the capacity to carry out large-scale missions. It automatically builds miniApp for users and helps to handle large tasks efficiently. A typical application scenario is that, when users want to find inspiration for weekends of family entertainment, Google searches can start from scratch and design a detailed programme of weekend activities。

GOOGLE AND SAMSUNG'S AI GLASSES ARE ON THE MARKET THIS AUTUMN

HARDWARE IS ALSO AN IMPORTANT THEME FOR THIS I/O SESSION。

Google's first public display of a new generation of AI smart glasses marked its return to the smart glasses market. Google announced that the company was working with Samsung and the glasses manufacturers Gentle Monster and Warby Parker to launch smart glasses with their own Gemini assistant. Google indicated that the glasses would be compatible with Android and iOS equipment and were expected to be listed later this year。

The director of Android XR products and platforms, Shahram Izadi, said: “This is just the first two designs of a larger series of products to be launched this fall.” He noted that the glasses were intended to provide information by “intimately broadcasting to the user's ears, not through screen displays.”。

Samsung stated that the device would combine the Android XR platform with Gemini AI's ability to create an "intelligible computing experience"。

The I/O Congress presentation showed that the AI glasses were capable of: real-time translation; identification of the surrounding environment; provision of navigation; voice interaction; and call to Gemini intelligence。

According to industry, Google is trying to replicate Android ' s ecological advantages in the age of smartphones by combining AI Agent with AR devices and cutting in advance to the next generation computing platform。

RECONSTRUCT AI SUBSCRIPTIONS: LAUNCH $100/MONTH DEVELOPER PLAN

IN ADDITION TO THE MODEL, ANOTHER IMPORTANT CHANGE FOR THE GENERAL ASSEMBLY WAS GOOGLE ' S REDESIGN OF THE AI FEE SYSTEM。

GOOGLE ANNOUNCED A NEW AI SUBSCRIPTION PLAN FOR DEVELOPERS AT $100 PER MONTH. THE PLAN WILL PROVIDE FOR:

• A higher amount of Gemini API calls
• AI Studio Advanced Functions
• Gemini Code Assist
• Google Cloud points
• Higher level Agent access rights。

Analysis suggests that this means that Google is beginning to commercialize its full-scale developers。

Until then, OpenAI and Anthropic had established pre-emptive advantages in the business and developer markets, while Google had a strong modelling capability but the developers were relatively slow to move ecologically。

Today, Google clearly wants to re-engage with developers through lower prices and deeper instruments。

AI Ultra reduced monthly fee from $250 to $200

Meanwhile, Google has adjusted its flagship AI subscription plan, AI Ultra。

While AI Ultra's monthly fee was $250, Google announced a downward adjustment to $200。

AI Ultra currently provides: top-level Gemini model access; Deep Think reasoning capability; advanced Agent functionality; AI video and image generation; AI programming capability; notebookLM advanced functionality; higher API and context limitations。

THE MARKET BELIEVES THAT BEHIND THIS PRICE ADJUSTMENT IS THE RAPID ESCALATION OF THE AI SUBSCRIPTION WAR。

During the past year, OpenAI launched ChatGPT Pro; Anthropic reinforced Claude Max; Microsoft expanded the Copilot subscription; and XAI is also promoting high-level Grok membership. Google’s downward adjustment of Ultra prices is both intended to increase the size of users and means that large model competition is moving from a “technology war” to a “price war”。