Google’s Gemini has undergone a rapid transformation since late 2025, moving from a conversational AI experiment into a system that is increasingly embedded across the company’s products and services. The pace of updates has been steady, but the cumulative effect is what stands out. Gemini today behaves less like a standalone chatbot and more like an AI layer that connects search, productivity tools, and mobile experiences.
This shift is already visible across Google’s ecosystem. Gemini is now capable of handling tasks directly inside its interface, from generating documents and presentations to supporting meetings and workflows, as seen in updates that brought deeper integration with Workspace tools.
A recent update reflects that shift in a practical way. Gemini now supports more natural continued conversations, allowing it to retain context across multiple prompts without requiring users to restate information. This builds on earlier work around memory and cross-app awareness, particularly inside Google Workspace, where Gemini can surface past interactions and use them to inform ongoing tasks.
Google has also been refining how users interact with Gemini across devices, including improvements to voice-based experiences and interface design. Recent updates have made conversations feel more fluid and less transactional, particularly in smart home and mobile environments where interaction needs to be immediate and intuitive.
That continuity changes how the system is used. Instead of issuing isolated prompts, users can now develop ideas, refine outputs, and move through tasks in a way that feels persistent rather than fragmented.
This update sits on top of deeper changes that have been rolling out since the end of 2025. Google’s newer Gemini models have introduced stronger reasoning, broader multimodal capabilities, and tighter integration with Search and Android.
Over the past few months, Google has expanded Gemini in ways that go beyond incremental improvements, focusing on how the system can be used across real workflows. The updates span productivity, personalization, and on-device AI, giving a clearer sense of how Gemini is evolving in practice rather than just in capability.
One of the more visible changes is how Gemini is being shaped into a productivity tool. Google has introduced features that allow users to organize their work more effectively, including the ability to structure conversations into dedicated projects and access notebook-style workflows directly on the web without a subscription. This builds on a broader push to make Gemini useful for ongoing tasks rather than one-off queries, especially as users begin to manage longer interactions and more complex inputs.
At the same time, Google is introducing ways to make Gemini more adaptable to individual users. Features like reusable prompt “skills” in Chrome allow users to save and reapply specific instructions, while improvements in image generation are moving toward outputs that reflect personal preferences by drawing on a user’s photo library. These changes point to a system that is not just responsive, but increasingly shaped by how each person uses it.
Since April, Google has also introduced newer models and more agent-like features. Gemini 3.5 Flash arrived in May, followed by Gemini 3.6 Flash in July, while Gemini Omni added conversational video generation and editing to the Gemini app. Google also introduced Gemini Spark, an early personal-agent feature designed to work in the background, and Daily Brief, which can use connected Gmail, Calendar, and Tasks information to organize a user’s day. Access to these features varies by subscription, region, and rollout status.
Some of the most important updates are happening at the system level. Google is continuing to push Gemini onto devices through its Nano models, while Gemini Intelligence is bringing more proactive features to Android. It is rolling out in waves starting with newer Samsung Galaxy and Google Pixel phones, with planned expansion to watches, cars, glasses, and laptops later in 2026. Alongside this, interface updates like the redesigned Gemini Live experience are making it easier to interact with the system in real time, particularly on mobile. These changes suggest a focus not just on what Gemini can do, but where and how it is used.
Google is also beginning to explore more sensitive use cases, including mental health support features, though these remain early and are positioned as assistive rather than standalone solutions. This reflects a broader pattern in how Gemini is being deployed, where new capabilities are being introduced cautiously in areas that require a higher degree of trust.
Google’s Gemini is Google’s generative AI platform, built around a family of models designed to process and generate text, images, audio, video, and code within a single system.
At its core, Gemini is a multimodal AI model. That means it can take different types of input, such as written queries, images, or voice, and interpret them together rather than in isolation. This allows it to understand context more effectively and produce responses that are more relevant and structured, especially in complex scenarios.
Gemini also represents a shift in how Google approaches AI products. Earlier tools like Google Assistant were built around commands and predefined actions, while Gemini is designed to handle open-ended queries, generate content, and support tasks that unfold over multiple steps. It is not just responding to requests, but interpreting intent and building responses that can evolve through interaction.
The platform itself has absorbed several of Google’s earlier AI efforts. Bard was rebranded as Gemini, while Duet AI became part of Gemini’s integration within Workspace. Today, Gemini exists both as the underlying model and the user-facing assistant, accessible through web browsers, mobile devices, and across Google’s ecosystem.
Recent model updates have further expanded its capabilities. Newer versions are better at structured reasoning, allowing Gemini to process more complex questions and provide answers that are more detailed and targeted. This is particularly noticeable in tasks that involve analysis, research, and multi-step problem solving, where the system is expected to go beyond basic responses.
Google Assistant and Google Gemini are built on fundamentally different approaches to AI, and that difference shapes how they are used.
Google Assistant was designed as a task-oriented voice assistant. It relies on predefined commands and integrations to perform actions such as setting reminders, controlling smart home devices, or answering simple queries. Its responses are typically tied to specific triggers, which makes it fast and reliable for routine tasks, but limited when it comes to handling open-ended or complex requests.
Gemini, by contrast, is built on large language models that can generate responses dynamically. Instead of relying on fixed commands, it can interpret intent, process different types of input, and respond in a more flexible way. This allows it to handle tasks that go beyond simple execution, including writing, summarizing information, analyzing inputs, and supporting multi-step workflows.
The difference becomes most apparent in how each system handles context. Assistant generally treats each request as a separate command, while Gemini can carry context across interactions, allowing users to refine queries and build on previous responses. This makes it more suitable for tasks that require iteration rather than immediate execution.
Google is now actively replacing Google Assistant across most supported mobile devices, as well as expanding Gemini to cars, watches, and Google Home speakers and displays. On supported phones, Gemini Intelligence can also automate tasks across apps and handle more system-level actions, although availability still varies by device, language, region, and rollout status.
Gemini’s capabilities continue to expand, but they are still defined by the environments in which the system operates. While it can generate content, analyze information, and guide complex tasks, it does not directly execute actions in the physical world. Google is exploring that possibility through its robotics efforts, where Gemini-like systems are being tested in real-world environments that extend beyond screens and devices.
In day-to-day use, however, the effectiveness of Gemini often comes down to how it is used. Because it relies on user input to guide responses, the quality and clarity of prompts can significantly influence the results. Learning how to structure queries and refine instructions is an important part of getting consistent, useful output from the system.
Gemini’s involvement in video creation has expanded beyond the planning and conceptual stages. It can take a rough idea and expand it into a structured script, suggest scene breakdowns, and help organize the flow of a narrative. Users can also generate short videos directly through Gemini Advanced, while Gemini Omni can create, remix, and edit videos conversationally using combinations of text, images, and existing video. It can change backgrounds, apply camera movements, and refine other elements through follow-up prompts. Access varies by plan and region.
Where Gemini becomes more useful is in iteration. Video production rarely follows a linear path. Scripts are rewritten, sequences are reordered, and ideas evolve as the project develops. With continued conversation, Gemini can retain the context of earlier drafts and feedback, allowing users to refine their work without starting from scratch each time. This makes it more practical as a collaborative tool rather than a one-time generator.
Processing information is still one of Gemini’s strongest capabilities, but the nature of its output has improved with newer models. The system is better at handling queries that require synthesis rather than simple summarization.
For example, when comparing multiple products, analyzing a topic from different perspectives, or breaking down a complex concept, Gemini can organize its response in a way that reflects the structure of the problem. This is a direct result of improvements in reasoning, which allow the system to handle multiple variables within a single response.
The difference is subtle but important. Earlier versions of AI systems often produced answers that were technically correct but lacked depth or structure. Gemini’s newer models are more capable of presenting information in a way that is easier to follow and apply.
Image generation within Gemini has become more flexible and iterative. Instead of relying on a single prompt to produce a final output, users can refine images through conversation, adjusting elements such as style, composition, and tone.
This approach changes how image generation is used. Rather than treating it as a one-step process, it becomes part of a broader workflow where ideas can evolve gradually. For designers and content creators, this makes the tool more adaptable, particularly when working toward a specific visual outcome.
Research is one of the areas where Gemini’s improvements in reasoning and context retention have the most impact. The system can handle tasks that involve gathering information, organizing it, and refining it over multiple steps, with newer capabilities allowing it to explore topics through structured, multi-step analysis, similar to how Google is positioning its deeper research tools within Gemini itself.
Google renamed NotebookLM to Gemini Notebook in July. It remains a standalone research product, but notebooks can sync with the Gemini app, and Google plans to integrate them more closely with Search. Newer versions can also execute code for source-grounded data analysis, although some advanced features remain restricted to particular subscriptions and Workspace plans.
This makes it useful for projects that require depth rather than quick answers. Users can begin with a broad topic, narrow it down through follow-up queries, and shape the output into a structured format without losing continuity. The ability to maintain context across these steps is what allows Gemini to support workflows rather than just individual queries.
Gemini Live introduces a more natural form of interaction through voice. Users can speak directly to the system, interrupt responses, and guide the conversation in real time, creating an experience that feels closer to a discussion than a sequence of commands.
Google has since integrated Gemini Live more directly into the main Gemini interface, allowing users to move between typing and speaking within the same conversation. It has also reworked microphone behavior so users can speak through longer ideas without being interrupted as easily.
The importance of this feature lies in accessibility and convenience. On mobile devices, voice interaction often feels more immediate than typing, especially when dealing with quick questions or ongoing tasks. Early hands-on experiences with Gemini Live also highlight how these interactions are evolving beyond simple voice commands into something more fluid and responsive.
One of Gemini’s defining characteristics is its ability to work across multiple types of input. It can process text, images, audio, and video within a single interaction, allowing users to combine different forms of information without switching tools.
This capability becomes more meaningful when paired with its integration across Google’s ecosystem. In Search, Gemini enables follow-up queries and deeper exploration of topics. In Workspace, it assists with documents and emails while drawing on existing context. These integrations change how users move between tasks, allowing them to carry intent across different environments.
A significant shift in Gemini’s development is its movement toward handling tasks that unfold over time. Instead of responding to individual prompts, the system is becoming more capable of supporting workflows that involve planning, execution, and refinement.
This is often described as a move toward agentic AI. In practical terms, it means the system can assist with tasks that require multiple steps, such as research projects, content creation, or problem-solving processes. The ability to retain context plays a central role here, as it allows Gemini to build on previous interactions rather than starting over.
Gemini Spark makes that shift more concrete. The cloud-based agent can continue working after a user closes a laptop or locks a phone, handle recurring tasks, connect with Google and third-party apps, and complete longer workflows under the user’s direction. Google later added integrations with services such as Tasks, Keep, Canva, and Dropbox, along with Chrome browsing that can use logged-in sessions with permission.
Daily Brief takes a narrower approach. Once enabled, it checks connected services such as Gmail and Calendar in the background and prepares a personalized morning digest with urgent updates, upcoming events, and suggested next steps.
Personalization is becoming a central part of how Gemini operates. By integrating with Google’s services, the system can draw on documents, emails, and past interactions to provide responses that are more relevant to the user.
This shift has important implications. A system that understands context can offer suggestions that are more aligned with individual needs, reducing the amount of input required from the user. At the same time, it raises questions about how data is used and managed, making user controls and transparency increasingly important.
Gemini is also influencing how information is presented. Instead of returning static blocks of text, the system is beginning to generate structured outputs that can be interacted with and refined.
These outputs might include organized summaries, multi-part responses, or formats that can be edited directly. The goal is to make the information more usable, allowing users to engage with it rather than simply read it. This represents a shift from AI as a source of answers to AI as a tool for building and refining content.
Google’s Neural Expressive redesign makes that idea more visible. Gemini can now construct responses using elements such as imagery, interactive timelines, narrated videos, and dynamic graphics rather than presenting every result as a conventional text response. The redesign has rolled out across the web, Android, and iOS.
Gemini’s integration into Search changes how users interact with information online. Instead of receiving a list of links, users can explore topics through follow-up questions, refining their queries within the same session.
This creates a more interactive search experience, where information is not just retrieved but shaped through conversation. It also reflects a broader change in how search engines operate, with AI playing a more active role in guiding users through information.
Google upgraded AI Mode in Search to Gemini 3.5 Flash in May, giving it a model designed to support longer reasoning processes, agent-like tasks, and more dynamic interfaces.
Google is extending Gemini into enterprise environments, where its ability to handle workflows and integrate with existing tools becomes more valuable. Businesses can use it to automate tasks, manage documents, and support internal processes.
This expansion reflects a broader strategy. Gemini is not just a consumer-facing product. It is being positioned as a platform that can operate across different scales, from individual users to large organizations.
The capabilities of Gemini are supported by Google’s underlying infrastructure, including custom hardware designed for AI workloads. These systems enable the processing of large and complex inputs while maintaining performance across different types of tasks.
Improvements in efficiency and scalability are critical to how Gemini operates. They allow the system to deliver faster responses, handle multimodal inputs, and support more advanced reasoning without compromising usability.
Gemini is available across a range of devices, including mobile phones, laptops, and desktop systems. On a PC or laptop, it can be accessed through any modern web browser via the Gemini website, offering a similar experience to other AI chat platforms. Google also offers a native Gemini app for compatible Macs running macOS Sequoia or later.
For those accessing the AI on a mobile device, Google Gemini is available through an Android app and on iPhones and iPads. The Android app currently requires Android 9 or later and at least 2GB of RAM, while the iOS app requires iOS 16 or later.
Android phone users can also switch from Google Assistant to Gemini and use it as their primary mobile assistant. Feature availability can vary depending on the device, account, language, and location, while some of Gemini’s more advanced system-level capabilities are limited to newer hardware.
Gemini also works with Samsung devices and can tap into native Samsung apps for information it might need. Google has also expanded Gemini to supported smartwatches, cars, speakers, and displays, although the available features differ between platforms.
Basic Gemini features can be used for free with a Google account. The free tier includes access to Gemini 3.6 Flash, varying access to Gemini 3.1 Pro, image generation and editing, Deep Research, Gemini Live, Canvas, and Gems, although usage limits apply.
Google AI Plus costs $4.99 per month in the US. It offers twice the usage limits of the free tier, access to additional features such as video generation and Daily Brief, 400GB of cloud storage, and Gemini inside some Google apps.
Google AI Pro costs $19.99 per month and increases Gemini usage limits to four times those of the free tier. It also includes higher access to Google’s more capable models, 1,000 Google Flow credits, Gemini in Gmail, Docs, Vids, and other apps, and 5TB of cloud storage. Google has previously offered a free trial of Google AI Pro, although trial availability can vary.
Google AI Ultra now starts at $99.99 per month for five times the usage limits of Pro. A $199.99 option raises those limits to 20 times the Pro allowance. Ultra includes the highest access to Gemini models and agent features, more Google Flow credits, at least 20TB of storage, and early access to features such as Gemini Spark and Deep Think in supported regions. These are US prices, and plans, benefits, and availability vary by country.
How you change Gemini’s settings depends on which device you use to access the AI.
In the Gemini mobile app, open the menu, tap your profile picture or initial, and select Settings. From there, you can adjust options such as Gemini’s voice, language, notifications, connected apps, personalization, and lock-screen access.
On the Gemini website, open the menu and select Settings & help. Some controls, including connected apps and activity settings, may appear in their own sections rather than inside one general settings page.
On Android, you can also manage whether Gemini acts as your device’s primary assistant. Open the Gemini app, tap the menu, select your profile picture or initial, and choose the option to switch between Gemini and Google Assistant where it remains available. The exact options may vary by phone manufacturer and Android version.
Gemini remains the more capable and broadly available generative AI assistant, but the comparison is no longer as one-sided as it was earlier in 2026.
Siri has traditionally focused on device controls and predefined actions, while Gemini is built to handle open-ended questions, generate content, process different types of media, conduct research, and support tasks that develop over multiple steps.
Apple announced Siri AI in June 2026, giving Siri more conversational capabilities, personal context, and the ability to take actions across apps. Apple had already been working toward a more personal assistant that could learn from information such as emails, after delaying its larger Siri AI upgrade. The newer Siri AI features are currently available for developer testing, with a public beta planned for later in 2026 on supported devices and initially in English.
That means Gemini still has the advantage for users who want those broader AI capabilities today. iPhone users can download the Gemini app while waiting for Apple’s newer assistant, particularly if Siri’s earlier limitations have already made them consider moving away from Apple’s ecosystem. The gap could narrow once Siri AI reaches more users, but its real-world performance will matter more than Apple’s announcement.
Google Gemini and ChatGPT are often compared because they solve similar problems on the surface, but they are built with different priorities, and that difference shows up in everyday use.
Gemini’s biggest advantage comes from how deeply it is integrated into Google’s ecosystem. It works across Search, Gmail, Docs, Android, and other Google products in a way that allows it to pull context from multiple sources and apply it within a single interaction. For users who already rely on Google’s services, this creates a more connected experience. Tasks like summarizing emails, working with documents, or refining search queries feel more seamless because the system is already embedded where the work is happening.
ChatGPT, by contrast, is less dependent on a single first-party ecosystem. That gives it flexibility across platforms and use cases, although it can now connect to external services and take actions through apps and plugins. It remains a strong general-purpose tool for writing, coding, research, image creation, and problem solving.
The difference becomes clearer when looking at how each system handles context. Gemini can draw directly on connected Google services and use that information within its workflows. ChatGPT also maintains context across conversations through memory and can reference other chats within projects, so it is no longer accurate to describe it as purely session-based. Users can control or disable that memory when they don’t want information carried between conversations.
There is also a distinction in how each system approaches reasoning. Both can handle research, analysis, content generation, and multi-step tasks, but Gemini is increasingly focused on tying those abilities to Google products and Android. ChatGPT’s approach remains more platform-neutral, with outside services added through its plugin and app system.
For multimodal use, both platforms support images, voice, files, and other formats. Gemini’s approach is more tightly woven into Google’s ecosystem, while ChatGPT keeps most interactions inside its own interface and connected services.
In practical terms, the choice comes down to how you plan to use the tool. Gemini is better suited for users who want an AI system that works directly across Google’s products. ChatGPT remains a strong choice for users who prefer a general-purpose assistant that works across a wider mix of platforms and tools, especially when comparing their capabilities in greater detail or testing them in real-world scenarios.
Microsoft Copilot enters the conversation from a different angle, particularly for users working within Microsoft’s ecosystem.
Copilot is deeply integrated into tools like Word, Excel, PowerPoint, Outlook, and Teams, making it particularly effective in structured, enterprise-heavy workflows. It is designed to enhance productivity within those environments, with a strong focus on documents, spreadsheets, meetings, and collaboration. Microsoft has also added agent-style capabilities that can perform multi-step actions directly within its productivity apps.
Gemini’s approach is similar in intent but broader in scope. It spans consumer and enterprise use cases, integrating across Search, Android, and Workspace. This gives it an advantage for users who move between personal and professional tasks within Google’s ecosystem.
The distinction here is less about capability and more about environment. Copilot feels most natural inside Microsoft’s suite of tools, while Gemini is designed to operate across Google’s platforms. Choosing between them often depends on which ecosystem a user is already invested in.
There is no single answer to this, because each system is optimized for a different kind of use.
Gemini stands out for its integration, context awareness, and ability to support workflows that span Google’s services and Android devices. It is evolving toward a system that feels embedded in everyday tasks rather than separate from them.
ChatGPT remains one of the most versatile general-purpose AI tools, particularly for writing, coding, research, media creation, and tasks that require access to a broader mix of external services. Its memory and plugin system also make it better suited to ongoing work than earlier versions were.
Copilot is strongest in structured Microsoft 365 environments, where integration with organizational documents, emails, meetings, and productivity apps is the priority.
Siri AI is being developed to compete more directly with generative assistants, but its newer capabilities are still in testing and are not yet as broadly available as Gemini. Its eventual strength will likely remain its integration with Apple hardware and apps rather than trying to mirror every feature offered by Google or OpenAI.
The more relevant question is not which one is better overall, but which one fits the way you work. As these systems continue to evolve, the differences between them are becoming less about raw capability and more about which services, devices, and workflows they can access.
Not long ago, Apple treated independent repair like an act of outright hostility. The Cupertino giant locked down parts with serial-number pairing, while repair tools remained firmly out of reach of anyone outside its authorized repair network. More importantly, the company waged a legal and lobbying battle against right-to-repair legislation.
I've covered enough of these fights over the years to know what a shift looks like. That's why what happened in October 2026 genuinely caught me off guard. The day the iPhone Duo went on sale, what may be the most mechanically complex iPhone to date, Apple pulled off a surprising about-face. It published teardown-style repair manuals for a handful of major components.
Apple's first foldable phone is now on shelves, in short supply, but with an unreal amount of buzz around it. After all, it's the first foldable iPhone, and there's plenty that this phone gets right. But there's also a lot that is either broken or underwhelming. What's peculiar is the dialogue around this phone, and how it has completely overshadowed every other attempt at making a phone with a similar form factor.
I've spent some time with the iPhone Duo, and I've used the Galaxy Z Fold 8 for almost three months now. In my opinion, Samsung deserves more credit. When the Galaxy Z Fold 8 arrived earlier this year, it was celebrated in some circles and ridiculed in others for supposedly copying what Apple would eventually deliver with the iPhone Duo. The dialogue was pretty odd, for multiple reasons. Samsung has been making foldable phones for years, so claiming that it copied a product Apple hadn't even launched was simply absurd.
For years, Samsung’s Galaxy A phones have been some of the easiest recommendations for anyone who wants to go the Android route without paying flagship prices. But that part of the lineup could soon become harder to find at a good price.
According to South Korea’s Electronic Times, Samsung is cutting Galaxy A production by about 20% in the fourth quarter compared to its original plans. Component orders have reportedly dropped by a similar amount, which could mean roughly 10 million fewer phones over the year. Samsung is also said to be considering another production cut of more than 10% next year.