ChatGPT and Google Gemini have both changed a lot in a very short amount of time. We are not talking about tiny visual tweaks that nobody asked for. These updates affect how you use voice AI, research, browser automation, files, projects, prompts, and even the way AI can complete tasks on websites for you.
ChatGPT has expanded access to its newer GPT 5.6 models, made voice conversations more useful inside projects, and added more ways to control reasoning speed and intelligence. Google Gemini, meanwhile, is pushing hard into agentic browsing with Gemini Spark, while Gemini Notebook is becoming a much more capable research and analysis workspace.
Some of this stuff is honestly insane. AI is moving beyond simply answering questions. It is starting to navigate websites, filter results, use browser sessions, organize research, execute code, and work toward goals with far less manual effort.
ChatGPT Live Now Works With Files and Projects
One of the most useful ChatGPT upgrades is the expansion of ChatGPT Live, also known as voice mode. You can now use voice conversations with files and within projects, which makes the feature much more practical for real work.
Instead of having a random voice chat disconnected from everything else, you can attach relevant material and keep the conversation inside a dedicated project. That matters because projects are where context starts to become genuinely useful.
You could create a project for:
- Career planning and job applications
- Trip planning and travel research
- Fitness tracking and personal goals
- Content ideas and writing workflows
- Client work, research, or ongoing business tasks
Inside voice mode, ChatGPT can now accept photos, files, camera input, and material from your device. That gives you a much more natural workflow. You can speak through a document, share a photo for context, ask questions about a file, and continue organizing everything in the right project.
For example, if you are building content ideas, you could keep your notes, uploaded files, and voice conversations inside one content project. If you are planning a trip, you could upload booking information, discuss your itinerary out loud, and keep everything together rather than bouncing between disconnected chats.
Choose the Voice, Language, and Intelligence Level
ChatGPT Live also gives you more control over how it responds. You can change the voice, set a language preference, and adjust the intelligence level.
The intelligence settings are especially interesting because they let you prioritize speed or depth depending on the task.
- Instant: Responds immediately. It is great when you want a fast back and forth, but it may not always deliver the most thoughtful answer.
- Medium: Takes a little more time to think before responding. This is a useful balance for everyday conversations and planning.
- High: Takes longer, but aims to give a stronger, more considered response. Use this for difficult questions, strategy, research, or work that needs more thought.
The easiest way to think about it is this: instant is the friend who answers before you finish your sentence, medium is the friend who pauses and thinks a little, and high is the friend who gives you that awkward silence before delivering a surprisingly good answer.
You can also leave language on auto-detect if you switch between languages or just want ChatGPT to identify the language naturally.
Within voice settings, there are model options including live, advanced, and standard. If voice mode is something you use constantly, the live model is the best fit for keeping conversations natural and helpful.
GPT 5.6 Is Now Available to Free ChatGPT Users
Another huge ChatGPT update is wider access to GPT 5.6 models. Free users now have access to GPT 5.6 Luna, while GPT 5.6 Sol has also been improved across the platform.
That means the gap between free and paid ChatGPT access has changed. Whether you use the free version, Plus, Pro, Enterprise, or Business, GPT 5.6 is now much more central to the experience.
GPT 5.6 Sol is designed to support a broad range of tasks, including:
- Quick answers from the web
- Web searches
- Planning and decision support
- Research assistance
- Complex questions that require deeper reasoning
For Plus and Pro users, the same model can power both quick responses and deeper reasoning. That creates a more consistent experience because you are not constantly guessing which model is best for one type of question versus another.
More Concise Answers and Better Reliability
One noticeable difference with GPT 5.6 is that its answers are intended to be more focused. Older instant-style responses could become extremely long, even when the question did not require a giant wall of text.
The new approach aims for answers that are more concise, more organized, and more useful. It also improves consistency between fast responses and deeper thinking, while placing a stronger focus on factual reliability.
Free users can also turn on Think when they want smarter, more deliberate responses. That is a big deal because it gives people more control over the tradeoff between quick answers and higher-effort reasoning.
More Control Over Models, Speed, and Effort
The ChatGPT interface has also changed how model and effort controls are displayed. Depending on the mode you are using, you may be able to choose between different response speeds and levels of effort.
Options can include:
- Fast responses
- Medium effort
- High effort
- Extra high effort
- Pro-level settings
In work-oriented settings, there is even more control, with GPT 5.6 Sol, GPT 5.6 Terra, and GPT 5.6 Luna available alongside speed and effort selections.
If you do not want to constantly test every setting, the simple recommendation is to use GPT 5.6. It offers the strongest overall balance of speed, accuracy, research capability, reliability, and agentic performance.
Gemini Spark Can Now Handle Browser Tasks
This is where things get really crazy. Google Gemini Spark has received a major upgrade that allows it to interact with Chrome and handle browser-based errands.
Gemini Spark settings now include controls related to remote browser history and remote execution. Make sure you review these settings carefully, especially if the AI is being connected to websites where you are logged in. Gemini Spark can potentially work with your browser session, accounts, passwords, and other sensitive information.
That is powerful, but it is also exactly why you need to understand what permissions are enabled before using it.
What Gemini Spark Can Do in Chrome
Gemini Spark is capable of navigating websites and completing actions based on natural-language instructions. You can ask it to search for information, use filters, collect results, fill in details, and work through browser tasks.
A travel-related request could look like this:
Find the cheapest nonstop flight from the Bay Area to Seattle for August 20 through August 24, then log in and fill in the information.
It can also be used for research and information gathering. For example, Gemini Spark was used to find an Eminem sneaker auction, sort sneakers by current bid from highest to lowest, and collect the relevant listings.
That required more than a basic search. The AI had to locate the auction site, interact with the site controls, change the sorting order to current bid from high to low, review the listings, and return the results.
That is the difference between a chatbot and an agent. A chatbot gives you an answer. An agent can move through steps, use tools, and complete part of the task.
Potential Gemini Spark Use Cases
- Research prices, listings, and product availability
- Gather and sort information from websites
- Schedule recurring checks for changing information
- Help complete browser-based forms
- Plan travel searches and compare options
- Create reusable skills for recurring online tasks
- Interact with websites that require multiple steps and filters
You can even schedule tasks. If you want a daily update on an auction, a listing, or a specific research topic, Gemini Spark can be set up to revisit the task. That is where this starts becoming incredibly useful for business workflows, personal research, and repetitive admin work.
AI agents are going to use the internet more and more. The important question is not whether that is coming. It is whether you know how to use these tools safely and effectively before everyone else does.
Better Prompts Still Create Better AI Results
Even with all of these model upgrades, browser agents, and new features, one thing has not changed: the quality of your output depends heavily on the quality of your prompt.
A vague request like, “Create a blog post about how MCP works in AI,” may produce something usable. But it will often be generic, poorly formatted, and annoying to turn into a polished piece of content.
A stronger prompt gives the AI the context and structure it needs to do a much better job. Instead of only giving it a topic, specify the outcome you want.
For example, define:
- Situation: What the content is for and who it should help
- Task: What you need the AI to create
- Objective: The desired result, such as a 1,000-word blog post
- Knowledge: Important details, constraints, or source material
- Format: Titles, headings, images, conclusion, or other requirements
A prompt optimizer can help turn a rough idea into a detailed instruction set. MyPromptBuddy, for example, can optimize prompts directly for Gemini, ChatGPT, Claude, Grok, and other large language models.
The difference can be massive. A basic prompt may create an unstructured draft. An optimized prompt can generate a complete article with a title, clear sections, stronger formatting, a conclusion, and suggested images that are easier to move directly into a website.
This is one of the easiest ways to get better results from AI without needing a better model. Give the model a better assignment.
Gemini Notebook Is Becoming a Smarter Research Workspace
NotebookLM is now being presented as Gemini Notebook, and it has received some serious upgrades. The core idea remains extremely useful: build a research workspace around your selected sources, then ask questions and generate outputs based on that material.
Gemini Notebook can now create documents and spreadsheets more directly, making it more useful for turning research into actual deliverables.
One of the biggest improvements is source control. You can choose which sources the notebook should use, then verify how many sources are selected in areas such as the data table.
This is important because source grounding is what makes a tool like Gemini Notebook valuable. Rather than pulling from everything and creating a mess, you can narrow the information available to the AI and keep the output tied to the materials that matter.
Native Code Execution for More Complex Analysis
Gemini Notebook is also gaining a secure cloud computer for each notebook. This allows it to write and execute code natively, helping with complex data analysis grounded in the sources stored within Google.
That is a meaningful upgrade for research-heavy work. It moves Gemini Notebook beyond summarization and toward analysis, especially when the task involves structured information or data that needs more than a simple written response.
This functionality is available for Google AI Ultra users and has been rolling out to Pro users on the web.
Gemini Is Becoming More Connected Across Google
Gemini Notebook is also being connected more closely with the broader Gemini experience. Cross-app syncing is coming between the Gemini app and the standalone notebook environment, with Gemini Notebook also expected to be brought into AI Mode in Search.
Google is clearly trying to make Gemini feel less like a separate tool and more like a layer across everything you already do.
Within Google, Gemini capabilities now include options to:
- Create images
- Ask questions about files
- Brainstorm ideas
- Create a canvas
- Upload images and files
- Choose between AI models
- Access AI Mode conversations and settings
Pay attention to AI Mode privacy settings as well. You can manage public links and review AI Mode history. Make sure nothing is public that you do not want associated with your name.
ChatGPT’s Chrome Extension Can Work Toward Goals
ChatGPT’s Chrome extension is also becoming much more powerful. With access to Chrome and the internet, it can work with files, folders, plugins, goals, planning tools, and skills.
The extension includes the same kind of model selection available elsewhere in ChatGPT, but the really interesting feature is the ability to set a goal.
For example, you could set a goal such as:
Understand all updates from the last seven days to ChatGPT, Gemini Notebook, and Google Gemini so I can create content about them.
From there, ChatGPT can begin loading tools, searching the web, determining what it needs to do, and pursuing the assigned objective. You can monitor its actions, pause it, edit the goal in the middle of the process, or delete it entirely.
It can also be used with plugins for tasks such as visualization, buyer research, and deal-related workflows. Add in plan mode, skill recording, files, and folders, and you have a much more capable browser assistant than a simple chat window.
Use the New AI Features With Intention
The biggest takeaway is simple: ChatGPT and Google Gemini are becoming more useful because they are becoming more connected to real work.
ChatGPT is improving voice workflows, projects, file handling, free model access, reasoning controls, and goal-based browser support. Google Gemini is expanding research capabilities through Gemini Notebook and pushing into web automation through Gemini Spark.
These tools are not perfect, and they should not be handed unlimited access without checking permissions. But if you use them thoughtfully, they can save a massive amount of time on research, planning, information gathering, content creation, and repetitive online tasks.
The people who get the most value from AI will not just be the people with access to the newest models. They will be the people who understand how to provide clear prompts, structure their projects, verify sensitive actions, and turn AI into a real part of their workflow.
Try one upgrade today: create a dedicated ChatGPT project, test voice mode with a file, optimize a prompt before sending it, or give Gemini Spark a small research task that does not involve sensitive information. The tools are getting crazier by the week, and the advantage goes to the people actually using them.
Frequently Asked Questions
What is the biggest new ChatGPT update?
ChatGPT has expanded GPT 5.6 access, including GPT 5.6 Luna for free users, while also adding stronger voice mode support for files and projects.
Can ChatGPT voice mode use files?
Yes. ChatGPT Live can work with uploaded files, photos, camera input, and project-based conversations, making voice mode more useful for ongoing tasks.
What can Google Gemini Spark do?
Gemini Spark can interact with Chrome to handle browser tasks such as searching websites, filtering listings, collecting information, completing multi-step web errands, and scheduling recurring checks.
Is Gemini Spark safe to use with logged-in websites?
Gemini Spark can work with browser sessions and connected accounts, so it is important to review remote execution, browser history, permissions, and privacy settings before giving it access to sensitive tasks.
How does Gemini Notebook improve research?
Gemini Notebook offers stronger source selection, document and spreadsheet creation, cross-app syncing, and native code execution for more complex analysis grounded in chosen sources.
Why should I use a prompt optimizer?
A prompt optimizer turns a vague idea into a structured instruction with context, objectives, requirements, and formatting details. This usually produces a more complete and useful AI response.
