ChatGPT has been a trailblazer among AI chatbots since its launch in 2022. And even though its most recent 5-series of models isn't quite as innovative as those that came before, the chatbot's replies still stand out for their accuracy and detail. Excellent complex reasoning and creative writing capabilities, alongside robust customization and memory, round out its feature set. However, ChatGPT can (and will) get things wrong from time to time, and it doesn't offer the productivity features of competitors. Ultimately, Google's Gemini is our Editors’ Choice winner among AI chatbots thanks to its superior value and dizzying array of features and integrations.

What Is ChatGPT?

Broadly speaking, ChatGPT is an AI chatbot you interact with via text or your voice. It can answer questions, do research, generate creative writing, handle math and science problems, make images, process files you upload, search the web, and much more. Features such as contextual understanding and memory, among others, differentiate ChatGPT from the earlier, simpler versions of Amazon's Alexa and Apple's Siri.

I find ChatGPT most useful for answering questions and doing research. Rather than tacking on “Reddit” to a Google search and scrolling through dozens of forum posts, I can often find what I need faster with ChatGPT. 

AI chatbots are imperfect technologies, though. If you look to them for information, keep in mind they can confidently get things wrong or make stuff up. I recommend checking sources outside of ChatGPT for anything mission-critical. 

How Does ChatGPT Work?

ChatGPT is like a complicated prompt-response equation with access to information on everything from basket weaving to quantum physics. The technology behind the scenes is primarily OpenAI’s 5-series of large language models that comprise artificial neural networks trained on massive datasets. ChatGPT doesn’t just rely on whatever data it was trained with, though, and can search the internet for up-to-date information.

GPT-5.2 is OpenAI’s flagship model series, and it’s just as good for casual conversations as complex reasoning; it automatically engages however much processing power it needs, depending on the complexity of the prompt at hand. However, you can also use GPT-5.2 Instant or GPT-5.2 Thinking, as opposed to GPT-5.2’s automatic mode, if you prefer manual control. You can use ChatGPT’s legacy models (GPT-4o, GPT-4.1, o3, and o4-mini, as well as various versions of GPT-5.1 and the original GPT-5) if you’re a Plus or higher-tier subscriber. 

As you use ChatGPT, you train its underlying models. In that sense, ChatGPT learns over time, and its performance isn't static. In other words, OpenAI doesn’t have to add new features or introduce new models to make ChatGPT’s responses more accurate, detailed, or relevant. However, this doesn’t necessarily mean ChatGPT won’t make the same mistakes as you interact with it or that you will always notice major changes from one week to the next.

Plans and Pricing: Not the Greatest Value

You can use ChatGPT for free, but paid plans for individuals, teams, and large organizations are also available, including Go, Plus, Pro, Team, and Enterprise.

The Free plan gives you limited access to GPT-5.2 Instant, not including GPT-5.2 Thinking. You also get limited access to features such as deep research, file uploads, image generation, memory, video generation, voice chat, and more. You will run into usage limits rather quickly with this plan, however.

Go ($8 per month) is ChatGPT’s introductory premium plan, and it increases usage limits beyond what you get for free, giving you more access to GPT-5.2 Instant, image creation, memory, messages, and uploads. You still don’t get access to GPT-5.2 Thinking or legacy models with Go, while some usage limits, such as for deep research, are the same as on the Free plan. OpenAI also notes that the Go plan might include ads.

Plus ($20 per month) is ChatGPT’s flagship plan. It adds features and increases usage limits across the board. Plus unlocks access to both GPT-5.2 Thinking and several legacy models, alongside custom GPTs, early access to new ChatGPT features, Sora 1 video generation, and more. For this review, I tested the Plus plan.

Pro ($200 per month) expands usage limits to the maximum and even allows for unlimited file uploads and image generation. It also offers exclusive features, such as access to the GPT-5.2 Pro model and previews of upcoming features. However, considering its prohibitive pricing, this tier just isn’t worth it for the vast majority of users.

For organizations, ChatGPT has Business ($25 per user per month, billed annually) and Enterprise (custom pricing) plans. These include integrations with various professional apps, enhanced security with admin controls, and support for compliance with privacy laws such as the GDPR, among other things. You can try the Business plan for free, but you need to contact OpenAI’s sales team directly to set up an Enterprise subscription.

ChatGPT is cheaper than some chatbots, such as Claude ($17 per month, billed annually) and Perplexity ($17 per month, billed annually). Still, ChatGPT’s Go plan is very limited and doesn't even provide access to GPT-5.2's complex reasoning. Moreover, Copilot's (starts at $10 per month) and Gemini's (starts at $8 per month) base tiers include advanced reasoning models and offer a wide range of features that ChatGPT doesn’t offer at any level, such as cloud storage. As such, ChatGPT's value is questionable.

Where Is ChatGPT Available?

ChatGPT is available on the web, though you can also download dedicated apps for mobile (Android and iOS) and desktop (macOS and Windows) devices. OpenAI also has an official ChatGPT extension for Google Chrome, which makes ChatGPT your default search engine. Official extensions aren't available for Firefox or Safari, but third-party ones are. ChatGPT has its own web browser, called Atlas, which you can use to access the chatbot.

Beyond ChatGPT’s apps and web interface, you can access the chatbot indirectly in lots of ways. For example, Siri can access ChatGPT to provide better answers. Microsoft Copilot uses OpenAI tech, including GPT-5.1 and GPT-5.2.

Many sites and services use OpenAI’s models, including Adobe Express and Perplexity. However, you won't always see explicit ChatGPT branding, and OpenAI likely doesn't take part in its development. To access the full suite of ChatGPT features, you should stick with an official version.

Getting Started and Interface: Easy to Use, Lots to Do

You can use ChatGPT without an account, but signing in unlocks many more features, including the ability to see your chat history and upload files.

ChatGPT’s dashboard is uncluttered. It presents an Ask Anything field front and center, with options such as dictation or web search. A menu on the left has tabs for Apps (ChatGPT’s integrations), Codex (an AI coding agent), your chat history, created images, custom GPTs, and projects. If you’re not sure what you can use ChatGPT to do, OpenAI conveniently places buttons like Analyze, Brainstorm, Code, and Create Image below the central search field. You can click these to see sample prompts.

(Credit: OpenAI/PCMag)

Although ChatGPT’s interface is standard for an AI chatbot, I miss some interface innovations from other chatbots. For example, Claude puts everything I generate, from apps to deep research, into an easy-to-access tab. It also generates creative writing in a sidebar that maintains line breaks and spacing across different displays and resolutions. These aren’t absolutely essential features, but I would love to see ChatGPT implement them.

You can ask ChatGPT pretty much anything, and responses usually generate quickly. It's easy to choose a model from a drop-down menu at the top left of the screen. The selection here varies depending on your plan, but ChatGPT’s primary models are GPT-5.2 Auto, GPT-5.2 Instant, and GPT-5.2 Thinking. Instant is best for quick responses that don’t require a lot of processing power, while Thinking is meant for more complex prompts. Auto decides how much time to spend on your prompt, flipping between Instant and Thinking as necessary. If your plan allows for access to ChatGPT’s legacy models, you can find those in this drop-down menu, too.

ChatGPT is generally more friendly and personable than Copilot or Gemini by default, but you can personalize it to suit your preferences. You can specify the name it uses for you and have ChatGPT remember certain information about you. You can also choose among eight personality options, ranging from the cynical to the professional. You can tweak specific characteristics of ChatGPT, too, including how enthusiastic and warm it is, and how often it uses emoji or includes lists in its responses. Finally, you can give ChatGPT specific custom instructions on top of all of the above. For example, I leave ChatGPT’s characteristics and personality at their default settings, but I give it custom instructions to push back (or even play devil’s advocate) whenever relevant and not to always agree with me. Other chatbots don’t give you this same level of customization.

ChatGPT can remember everything you tell it and reference prior conversations. It's a nice upgrade to the chatting experience, making it easy to pick up where you left off across different chats. Other chatbots, such as Copilot and Gemini, have memory abilities, but ChatGPT’s is the most comprehensive I’ve tested. For example, it can tell me what the first message I sent it was, whereas Copilot and Gemini can’t.

The buttons below responses allow you to copy, read aloud, or regenerate responses. To share a whole chat, click the Share button at the top right of the interface. Deep research and generated images also get shareable links. ChatGPT has a relatively unique button at the top right of its interface, the Temporary chat button. This allows you to have chats that it won't save to your chat history; it's a nice option for sensitive or out-of-character topics.

Voice Chat: Lifelike But Imperfect

You can find ChatGPT’s voice chat and speech-to-text functionality on the right side of its central text field. Voice chat works in ChatGPT's web interface, whereas Gemini limits this functionality to its apps and the Chrome browser. Getting started with voice chat is easy, and you can talk to ChatGPT like you would to another person, with no need to press enter after speaking. You can choose from a variety of different voices and mute your microphone with a single click when you aren’t speaking, too. All the customizations I mentioned above also apply to ChatGPT during voice chat, so changing its voice is just the beginning of what you can adjust.

Beyond just chatting, you can share your camera or screen with ChatGPT on mobile, so it can see what you’re seeing. Similarly, you can also send ChatGPT photos during voice chats. This functionality is as good as ChatGPT’s image recognition, which means it’s generally quite skilled at parsing visual information. Still, it’s not flawless and will misidentify things from time to time.

In general, ChatGPT’s voices sound lifelike and are about on par with those in Copilot. But they aren’t flawless. Sometimes (and especially on mobile), for example, the voices distort. And they don’t sound as natural as those in Sesame, which inflect more realistically and match tone better. Nonetheless, ChatGPT’s voice chat is robust and capable, so it’s just as easy to communicate via voice as via text. 

Web Search: Detailed Responses

Searching the web is a standard feature of AI chatbots. Whether it’s ChatGPT (GPT-5.2 Instant), Copilot (Quick Response), or Gemini (3 Fast), all the chatbots I tested answered most questions correctly, such as when the next season of Plubirus would come out and the latest news on Donald Trump and Greenland. ChatGPT and Gemini were even able to tell me what Warframe’s current Incarnon weapon rotation was at the time of writing, which is a question chatbots often fail to get right, Copilot included. 

Response detail and length varied, depending on the bot and the prompt, but it was generally consistent. ChatGPT generated responses slightly faster than Gemini and significantly more so than Copilot. I also appreciate how ChatGPT pops out articles when you hover over linked sources with your cursor, including names, titles, and publication dates. Gemini lacks this feature almost entirely, but Copilot does something similar, albeit without showing publication dates and including a source’s web address instead of an excerpt. ChatGPT also, when relevant, includes article tiles for further reading on a topic, which the other bots don't. However, ChatGPT, unlike Copilot and Gemini, doesn’t highlight relevant text in its responses when you hover over a linked source.

ChatGPT automatically includes images in responses whenever relevant. Copilot doesn't do this as often, while Gemini won't unless you ask it to. Anecdotally, while all three chatbots include tables in their responses, ChatGPT tends to do so less often, even when a table would be the cleanest way to present a particular piece of information.

Deep Research: High-Quality Reports With Missing Features

AI chatbot deep research is, simply put, awesome. This feature lets you ask a question or pick a topic for ChatGPT to research and generate a report on. In my experience, these reports often run dozens of pages, with upward of 50 or 60 sources.

I compared ChatGPT’s, Copilot’s, and Gemini’s deep research features across a variety of prompts, including comparisons of two different espresso machines, a gear progression guide for my Ironman account in Old School RuneScape, which hairstyling products would be a good fit for my hair, and more. 

ChatGPT and Gemini consistently took longer to generate reports than Copilot, but their reports were longer and more in-depth. Gemini always read through the most sources, but it ended up using a similar number of sources as ChatGPT and Copilot. The information across these reports was generally consistent, but I much prefer the longer reports from ChatGPT and Gemini, since I want the most comprehensive coverage possible when I conduct deep research.

ChatGPT lacks some quality-of-life features here. For example, it doesn't highlight relevant text when you hover over a linked source like Copilot. Gemini has a menu at the top of its reports that lets you quickly jump between sections, and it also makes it easy to generate flashcards, infographics, quizzes, and web pages based on reports with a single click. Furthermore, Gemini allows you to export reports to Google Docs, and Copilot does the same with Microsoft Word. ChatGPT doesn’t do any of that. I also prefer Gemini’s deep research interface, which opens reports in a large sidebar. This makes it easier to read reports in the context of your overall conversation.

The tone of the reports differs significantly. Copilot's and Gemini's reports are somewhat formal, while ChatGPT’s read more like elaborate Reddit posts. I find ChatGPT’s reports more engaging, but you might disagree.

Image Generation: Close in Quality to Gemini

You can generate images with ChatGPT, just as you can with Copilot or Gemini. To start, I tested the chatbots' abilities to create photorealistic images. I used the following prompt in ChatGPT (GPT Image 1.5), Copilot, and Gemini (Nano Banana Pro): "Generate me an image of a cozy suburban home with an open floor plan. I want to see a nice living space with a dining room, kitchen, and living room. Nothing too fancy." ChatGPT's (first slide), Gemini's (second slide), and Copilot's (third slide) results are below:

ChatGPT performed well, producing a believable apartment with no significant issues. Copilot produced something similar, but its image was less detailed than ChatGPT’s, giving it a stock-photo feel. Gemini’s image is my favorite overall because it's slightly more detailed than ChatGPT’s.

Next, I asked the chatbots to make comics with the following prompt: “Generate me a six-panel comic image of a high fantasy kingdom where wizards have replaced spellcasting with PowerPoint presentations. Make sure there’s a major twist in the final panel.” Take a look at ChatGPT's (first slide), Gemini's (second slide), and Copilot's (third slide) comics below:

ChatGPT’s comic is coherent and detailed, with no major issues. However, Gemini’s comic stands out for its more robust text generation, which is almost amusing, such as its “Ethereal Connection” lag. At the other end of the spectrum, Copilot’s comic tells an incoherent story, lacks a twist, and uses an overly simplistic art style.

For my last test, I asked the chatbots to generate a diagram. My prompt was: "I've got an Ethernet switch, a modem, and a router. I want to connect my PC, PlayStation, smart light hub, and smart TV via Ethernet. I want the modem to connect to the router, and then the router to connect to the switch. Then, I want the switch to connect to the devices. Generate me a diagram showing this." ChatGPT's (first slide), Gemini's (second slide), and Grok's (third slide) results are below: Check out ChatGPT's (first slide), Gemini's (second slide), and Copilot's (third slide) diagrams below:

ChatGPT’s and Gemini’s diagrams are nearly flawless, but they still have some small issues. For example, ChatGPT includes a strange connector to the right of its Ethernet switch, and Gemini’s arrow near its PC connection is backwards. Copilot’s diagram is missing a connection between its modem and router and uses a simplistic art style, but it’s otherwise serviceable.

Image Editing: Not Quite the Best

If you edit an image with AI, you don’t want it to look AI-generated, which is a common problem. Does ChatGPT fall victim to this, or does it impress, like Gemini’s Nano Banana, which earned a Technical Excellence award? To find out, I gave ChatGPT (GPT Image 1.5), Copilot, and Gemini (Nano Banana Pro) two editing challenges. For the first test, I tasked them with removing a pesky hand from a landscape shot. ChatGPT’s (first slide), Gemini's (second slide), and Copilot's (third slide) results are below:

Similar Products

Our Current Picks for The Best AI Chatbots for 2026

Google Gemini

4.0 Excellent

Read Our Review

ChatGPT

4.0 Excellent

Read Our Review

Microsoft Copilot

4.0 Excellent

Read Our Review

Perplexity

3.5 Good

Read Our Review

Claude

3.5 Good

Read Our Review

Grok

3.5 Good

Read Our Review

All chatbots had issues with this task. The colors in ChatGPT’s image aren't authentic to the original, and a tessellation effect across fine detail is distracting. Copilot’s image matches the original's colors best, but it looks fairly grainy. Gemini’s image has grain problems, too, but it's the highest-resolution image of the group and the only one that maintains the original’s aspect ratio.

Next, I gave the chatbots a picture of my kitchen, alongside pictures of chairs and a table. Then, I asked them to replace the chairs and table in my kitchen with those in the images I provided, matching the lighting (and aspect ratio) of the original as much as possible. Check out ChatGPT’s (first slide), Gemini's (second slide), and Copilot's (third slide) results below:

ChatGPT did a good job overall, but it distorted some background details, and its lighting around the chairs and table isn’t impressive. Gemini's image also shows some background distortion, but, once again, it is the only one that matches the original's aspect ratio. Furthermore, it's the highest resolution of the three, and the lighting on its chairs and table is the most realistic. Copilot's image shows the most distortion.

ChatGPT integrates with Adobe Express and Photoshop, allowing you to leverage some of those apps' functionality. For example, you can select Adobe Photoshop in ChatGPT’s central text field drop-down menu and prompt it to adjust an attached image’s contrast. ChatGPT then overlays a contrast slider over your image so you can adjust it. I appreciate how straightforward this all works, but it's just as easy to ask ChatGPT (or another chatbot) to increase or decrease an image’s contrast. In other words, you still need dedicated photo editing software for advanced editing.

Video Generation: Occasionally Impressive

AI video generation is becoming mainstream, thanks in part to ChatGPT’s Sora AI. The old version of Sora, Sora 1, is still available to paid ChatGPT subscribers, but the current version, Sora 2, is rolling out to a wider audience after being invite-only initially. If you don't have access to it yet, you will likely have it soon.

Sora can create videos based on just about any prompt, and its videos also have audio. Results sometimes seriously impress, but they can also disappoint, especially without careful prompt calibration and multiple iterations. Sora isn’t just a video generation model, either: It's also a TikTok-like social platform for sharing AI videos.

To evaluate their video generation abilities, I gave ChatGPT (Sora 2) and Gemini (Veo 3.1: Quality) three prompts, starting with: "Somebody going about their daily life in a trendy apartment with rustic decor." ChatGPT’s video doesn’t impress. It treats a levitating cup as if it were a pour-over, and afterward, the person in the video awkwardly crouches in front of a table. Veo’s video isn’t great either. In it, a person cooking grabs a spoon, but the spoon he grabs duplicates, leaving one on the table and one in his hand. Oddly, a record player is also on the kitchen counter. In both videos, the audio is slightly distorted, doesn’t sync up perfectly, and is missing certain sounds.

To test the chatbots' abilities to handle complex motion, I asked them to create a video of somebody solving a Rubik’s Cube in a competitive setting with the following prompt: “Show me a pro Rubik's Cube solver solving a cube.” Once again, neither video is especially good. Both feature distorted cubes, which means the audio in both videos doesn’t quite sync with what’s on screen. ChatGPT’s timer doesn’t make sense, while Veo’s camera zoom is distracting. The voice of ChatGPT’s persona also has a slightly distorted quality, making it feel AI-generated.

My final test was for text generation within a video: "Generate me a video of a teacher in front of a class writing down y = mx+b on a whiteboard while explaining the concept." Unsurprisingly, there are significant issues with these videos as well. ChatGPT’s text is nonsense, and the voice of its teacher is, again, distorted. Veo’s video, confusingly, starts with “y = __ +b” on a whiteboard, and the teacher fills in the “mx” portion, while most of what its teacher actually says is garbled nonsense. Neither delivers on my prompt.

Even though my tests suggest otherwise, you can generate impressive videos with both ChatGPT and Gemini. However, this requires numerous prompt tweaks, multiple generations, and considerable time. If you pay for ChatGPT’s expensive, $200-per-month Pro subscription, you can use Sora 2 Pro, rather than Sora 2. I ran the same prompts through Sora 2 Pro below, and although the generated quality seems slightly higher, the videos still feature various errors and distortions. 

ChatGPT Pro also lets you leverage Sora 2's Storyboard feature, which breaks down videos into individual scenes that you can script. These videos can be 25 seconds in length compared with 10 seconds for the standard ones. Although this feature is useful for generating more complex videos, it doesn't result in meaningfully fewer errors and distortions in testing. Veo has a somewhat similar Flow tool for editing videos and stitching them together, but it doesn't avoid issues either.

Uploading and Processing Files: Deep Document Analysis

You can upload files to ChatGPT, including a resume that needs critiquing, text that needs translating, or something else entirely. Uploading a file seems simple, but processing images and understanding documents is actually quite complicated.

As a test, I asked ChatGPT (GPT-5.2 Thinking, Extended Thinking), Copilot (Smart Plus), and Gemini (3 Thinking) to identify as many components in my computer as possible with as much specificity as they could muster. I provided a picture of my computer with the glass side panels attached, so the chatbots needed to deal with the reflections.

(Credit: Ruben Circelli)

ChatGPT performed well, identifying my Aqua Computer D5 Next pump, ASUS ROG Hero motherboard, and black G1/4” compression fittings, among other things. Gemini also did very well, spotting my Aqua Computer Leakshield, G.Skill Trident Z5 RGB DDR5 RAM, Lian Li O11 Dynamic EVO XL case, and more. Copilot struggled here, failing to provide any specifics.

Moving on to document analysis, I prompted ChatGPT (GPT-5.2 Thinking, Extended Thinking), Copilot (Think deeper), and Gemini (3 Thinking) to read the manuals for my Leakshield (a PC water-cooling device), my motherboard, and my motherboard’s BIOS. I asked them to tell me, based on the provided manuals, whether I needed Windows to use the Leakshield, whether my motherboard supported USB pass-through power, and how to enable it if so.

All three chatbots gave me the correct answers: I didn’t need Windows, my motherboard did support pass-through power, and enabling it was a BIOS toggle. That said, both ChatGPT and Gemini referenced specific pages from the manuals in their answers, whereas Copilot didn’t, even though its answer came from the manuals, too.

When it comes to recognizing images, even ChatGPT and Gemini made mistakes, and I’ve seen AI chatbots hallucinate sections and make up quotes from provided documents. You shouldn't use chatbots for anything important, but they are fine for lower-stakes tasks.

Creative Writing: Handles Tough Prompts

Chatbots can tell jokes, write emails, and generate just about any text you can imagine. But as chatbots become more advanced, judging their creative writing mettle requires more than evaluating whether they can tell a coherent story. 

I gave the following prompt to ChatGPT (GPT-5.2 Instant), Copilot (Quick Response), and Gemini (3 Flash): "Without referencing anything in your memory or prior responses, I want you to write me a free verse poem. Pay special attention to capitalization, enjambment, line breaks, and punctuation. Since it's free verse, I don't want a familiar meter or ABAB rhyming scheme, but I want it to have a cohesive style or underlying beat." 

ChatGPT wrote the best poem based on my instructions. It has the largest variety of punctuation, reads the least like prose, and avoids any glaring issues. It's the only one with a title, too. Copilot’s poem largely meets the criteria in my prompt, but it doesn’t stand out in any particular way. Gemini's poem reads the most like prose of the three, which doesn’t say much for its enjambment, line breaks, and punctuation. 

Complex Reasoning: Math and Science Skills

My test of complex reasoning spans computer science, math, and physics. I gave ChatGPT (GPT-5.2 Thinking, Extended thinking), Copilot (Think Deeper), and Gemini (3 Pro) exam questions from undergraduate courses at Harvard, MIT, and Stanford. Then, I compared their answers with the solutions. I gave the chatbots these questions in PDF format, and I told them all not to search the internet for solutions.

ChatGPT performed the best, getting only one physics question wrong. Gemini was a close second, getting just one math and one physics question incorrect. Copilot missed four math questions and two physics questions, putting it at the bottom of the list.

All these chatbots can assist with homework, and ChatGPT even has a dedicated study mode that helps you learn new things rather than just giving you answers. Still, I don't recommend trusting them blindly when anything important is at stake. Make sure to verify the answers you get by checking them against the solutions.

Custom GPTs, Shopping, and Integrations: Situationally Useful

Custom GPTs are integrations you can prompt as you would ChatGPT itself. You can find custom GPTs from OpenAI or third parties, such as Canva or Wolfram Alpha. They’re single-use versions of ChatGPT, such as for designing logos with the Canva GPT or solving math equations with the Wolfram GPT. Gemini’s Gems are similar to some of OpenAI's custom GPTs, like the Creative Writing Coach custom GPT and the Writing Editor Gem, but there are far more custom GPTs than Gems. Third parties also can’t create and publicly share Gems like they can custom GPTs.

While the concept of custom GPTs is intriguing, the results don’t impress. For example, I asked the Canva custom GPT and ChatGPT to create a new logo for PCMag, and both followed my instructions. I do like that I can open the custom GPT logo in Canva by clicking it, but other than that, there isn’t much difference between the two.

Shopping is another ChatGPT feature. When you ask for buying advice, ChatGPT gives you a list of clickable, scrollable tiles with products, followed by descriptions of each entry. If you click on a product, a sidebar appears with links to retailers. Compared with Copilot and Gemini’s equivalent shopping features, ChatGPT’s sidebar has less information; you won't find pros and cons, specs, and user reviews.

(Credit: OpenAI/PCMag)

The buying recommendations themselves are hit-or-miss. I asked ChatGPT for the best laptops of 2026, but it didn’t recommend many I expected to see. ChatGPT also simply lacks competitors' features. For example, Copilot lets you buy certain products directly, while Gemini has a unique virtual try-on tool for clothing. You shouldn't trust a chatbot with financial decisions in general.

Gemini integrates with Google apps, such as Gmail and Docs, while Copilot can do the same with Microsoft 365 apps, like Outlook and Word. ChatGPT doesn’t benefit from those close connections, but it does offer integrations. As mentioned, you can access some of Adobe Express’s and Photoshop's functionality within ChatGPT, and the same is true for other services, too. For example, after I connected ChatGPT to my Instacart account, I had no trouble getting the bot to add ingredients from a recipe to my cart. ChatGPT works with a long list of different apps and services, from Google Drive to Spotify, so you will likely find some use for its integrations.

Agent: Just a Curiosity for Now

In the vein of Google’s Project Mariner, ChatGPT offers an Agent mode that lets it act as a virtual assistant. Agent operates in a dedicated, real-time web browser instance within ChatGPT’s traditional interface. It can handle various tasks, such as ordering groceries, planning a vacation, and much more. In Agent mode, ChatGPT is doing something for you, rather than just providing information. If you want to try out the Agent feature, you need to subscribe to ChatGPT’s Plus tier or higher.

(Credit: OpenAI/PCMag)

My experience with Agent is pretty much the same as when I use the agents of AI web browsers: disappointing. Agent usually takes longer to do a task than I can myself and frequently runs into issues. It might, for example, get stuck at a specific step in its process or be unable to access certain sites. When I asked Agent to find an oyakodon recipe and add the relevant ingredients to my Instacart order, it took a staggering 15 minutes. Moreover, all it did was grab the first recipe from a Google search, rather than putting any meaningful effort into finding an especially authentic or high-quality recipe. Even within ChatGPT, I can just call on its Instacart integration, provide a link myself, and have it add the ingredients to my cart instantly. 

Yes, Agent can complete some tasks, but it needs to get faster and more reliable before I’d ever seriously consider using it to save time. Of course, you might find a more meaningful application for Agent. For what it’s worth, it’s better than some AI agents I’ve tested, such as Project Mariner, which often wasn’t able to complete tasks at all.

Atlas Web Browser: It Doesn’t Stand Out

ChatGPT has an AI web browser, Atlas, which you can download and use for free. However, Atlas is available only on macOS at the time of writing. Another important caveat is that while you can use Atlas for free, its AI agent feature (the defining feature of AI web browsers) requires a premium ChatGPT subscription.

Atlas didn't impress in testing. It doesn’t have the quality-of-life features of traditional browsers, its AI features aren’t particularly unique, and the Agent mode doesn't work well. If all you want is a web browser with some AI functionality, and you really like ChatGPT, there’s no harm in downloading it, but don’t expect a life-changing experience.

What Can’t ChatGPT Do?

Though this may surprise Gen Z, ChatGPT can’t be your friend, romantic partner, or therapist: It’s not conscious. Whether the movie "Her" excites or terrifies you, ChatGPT isn’t there yet. Not only does it lack sentience, but there are a variety of restrictions on what else ChatGPT can do.

Adult content, anything illegal, realistic images of people, and taboo subjects like hate speech are against OpenAI’s policies. ChatGPT does try to avoid these, but it’s fairly easy to slip through the cracks and get responses that violate policy. In my testing, ChatGPT’s filters were much weaker than Copilot’s, but they weren’t as lax as Gemini’s and Grok’s, which don't restrict much at all.

ChatGPT has practical limitations, too. The context window, which ranges from 128K in the Pro and Enterprise plans to 196K for GPT-5 Thinking, determines how much data ChatGPT processes at once, including your prompt, ChatGPT’s response, and relevant contextual information. Depending on your plan, you might need to break long prompts and complex requests into multiple messages. ChatGPT’s context window pales in comparison to the 1M token context window you get with Gemini. Anecdotally, though, I didn’t run into context window caps during testing.

OpenAI caps usage across plans, except for the Business and Pro plans, which offer unlimited usage. However, caps are often dynamic. Free users can send up to 10 messages every five hours, while Plus users can send up to 160 messages within a three-hour window, but this may change during peak usage hours. With GPT-5 Thinking, you can send up to 3,000 messages per week. In my experience, I didn’t hit usage limits with the Plus plan, but I did easily with ChatGPT's free plan.

Is Your Data Safe With ChatGPT?

Reading through ChatGPT’s privacy policy, OpenAI collects a significant amount of data. This includes account information, any information you provide in surveys or events, and communications with OpenAI. Beyond the basics, OpenAI also collects a host of analytics data, including location information from IP addresses or GPS, information about your devices, log data from your browsers and devices, and usage data.

Perhaps most importantly, OpenAI collects user content, including whatever you write in prompts or send in messages to ChatGPT. The purpose of this is to further train ChatGPT. OpenAI takes “steps to reduce the amount of personal information in [its] training datasets before they are used to improve and train [its] models,” but you can opt out.

OpenAI enables user content data collection for training use by default. Although that collection doesn't include audio or voice recording data by default, you can choose to allow it if you wish. Despite these collection policies, OpenAI never sells or shares personal data for advertising purposes. That said, what you tell ChatGPT can show up in court, so make sure to think twice before discussing any crimes.

Someone hacked OpenAI in 2023 but reportedly didn't get access to OpenAI’s core systems. OpenAI was not forthcoming with this information, though, as the news broke only in 2024. OpenAI has continued to be a target for hackers, including when Chinese hackers sent malware to OpenAI employees. Given how much data OpenAI collects, its prior breaches, and its inconsistent reporting of those breaches, I don’t recommend sharing anything too sensitive with ChatGPT.