AI browsers in 2026: Arc, SigmaOS, Perplexity Comet — are any worth switching?

Arc sits frozen under Atlassian. SigmaOS now charges for features Arc gave away. Comet dropped its $200/month paywall and went cross-platform. Once you know where to look, the AI browser market finally has a clear answer. Every few months, someone in my network shares a screenshot of a shiny new browser setup and wonders aloud why everyone isn't using it. There's always a tidier sidebar, a cleverer tab system, or an AI model tucked into the address bar. The replies split just as predictably. Half come from enthusiastic converts. The other half come from people who gave it a try, ran into a wall around their third Chrome extension that refused to work, and quietly went back. The 2025-2026 wave of AI browsers has pushed that debate into serious territory. Perplexity cut Comet from $200 a month to free. Atlassian paid $610 million for The Browser Company, then left Arc frozen while the team started over. SigmaOS kept shipping, OpenAI launched Atlas, and half a dozen others declared themselves the future of browsing. The question in the title finally has a real answer. Let's take the contenders one at a time. Arc: beautiful, frozen, and no longer what it was Arc earned its loyal following by getting something important right. The tab bar, a metaphor inherited from Netscape in 1994, was never built for someone juggling 40 open tabs, six research threads, and three client contexts. Arc's sidebar, Spaces, and split view turned the browser into a workspace instead of an overstuffed desk drawer. For the first time in years, a browser felt like it had been designed by someone who used one for real work. Then the company stopped building it. The Browser Company paused all active Arc development in May 2025 and moved its focus entirely to Dia. Atlassian acquired the company in September 2025 for $610 million. Arc still downloads, installs, and runs on Mac and Windows as of June 2026. The latest macOS build, v1.152.0, shipped on June 18, 2026 on Chromium 149. It gets security patches and nothing else. The team that built Arc now works on Dia under Atlassian's ownership, and the acquisition announcement mentions Arc just once, in passing. So should you install Arc today? No. Anyone setting up a new machine or starting a fresh workflow shouldn't build around a browser with no development team and an uncertain future for its sync infrastructure. The sidebar will still feel wonderful, but the product slips further behind its competitors every month. If you're already on Arc and your workflow depends on it, there's no urgent reason to leave, since it still works well. Even so, start planning your exit while you still have choices. That beats being forced into a rushed move by a sync failure or a security gap. Dia: what Arc's team built next, and what it isn't Dia opened to the public on macOS on October 9, 2025. It comes from the same team behind Arc, now backed by Atlassian's enterprise resources. Anyone hoping for "Arc 2.0 with AI" will be disappointed, and early-2026 reviews agree on this point: Dia is a different product, built for a different audience. Arc served power users who wanted a heavily customized browser organized around their workflow. Dia targets enterprise teams that want AI assistance woven into a fairly conventional browsing experience. Its URL bar also works as an AI chatbot, so you can summarize tabs, upload files for analysis, and use Skills, which are custom AI shortcuts that act on whatever page or tab group is open. Skills is the best thing about Dia. You can save one prompt for reviewing GitHub pull requests, another for pulling key claims out of academic papers, and a third for summarizing an email thread, and each runs on your current context with a single keystroke. There are real limitations. Dia needs macOS 14 or later on Apple Silicon (M1 or newer), and there's no Windows version yet. Arc-style Spaces, which let power users separate their work contexts, are missing. Early-2026 updates added a sidebar mode and tab groups that point in that direction. In early 2026, one Reddit commenter called Dia "Chrome with a ChatGPT sidebar," which is harsh but not entirely unfair for a launch-stage product. It's improving weekly. Whether it will ever satisfy former Arc users is a separate question from whether it's a capable AI browser for people coming to it fresh. If you're judging Dia without any Arc nostalgia and you own a compatible Mac, the Skills feature alone justifies a test run. That workflow idea is genuine, and nothing else offers it in the same form. SigmaOS: the Arc lookalike with a funding problem SigmaOS is for people who want the nearest thing to Arc's layout philosophy from a team that's still actively shipping. It's Mac-only, backed by YC, and built around workspaces, a Command Bar, and layers of productivity features that overlap heavily with what Arc pioneered. If you used Arc Spaces, SigmaOS Workspaces will feel familiar within an hour. The trouble with SigmaOS isn't the product. It's the math. Arc raised more than $128 million across several rounds and still couldn't find a sustainable business model. SigmaOS has raised roughly $2 million, and it competes in a market where Google spent $20 billion simply to be Safari's default search engine. Building and maintaining a browser is brutally expensive per user at any funding level, and at $2 million, SigmaOS has a runway an order of magnitude too short for browser-scale infrastructure. The pricing makes things worse. SigmaOS costs $20 a month, and Arc gave away everything SigmaOS charges for. That comparison isn't entirely fair, since Arc had no revenue and no future, which is how you can afford to give things away. Still, from the user's side, paying $20 a month for a browser is a bet that the company will survive long enough to justify the cost of switching. Given the funding gap, I can't recommend that bet. SigmaOS built a good browser, so the product isn't the problem. A company with $2M in funding, up against a market where Google pays $20B a year just for default status, has a structural problem that good software can't solve. If you're already using SigmaOS and happy with it, enjoy it, but keep watching the company's finances. As a new install, it doesn't clear the bar on risk versus value when free alternatives exist. Perplexity Comet: the AI browser that earns the switch Comet began in July 2025 as a $200-a-month product that only a particular breed of research-obsessed power user would consider. On March 18, 2026, Perplexity removed the paywall worldwide and released Comet on macOS, Windows, iOS, and Android at the same time. Its competitive position changed overnight. Comet swaps the URL bar for a prompt box. Type a URL and it navigates. Type a question and Perplexity answers with citations. Type a task and the AI gets to work, so the line between browsing and researching disappears. Got fifteen tabs open on one topic? Ask Comet to synthesize them into a single summary. Shopping around? Have it compare prices across retailers and surface discount codes. Reading a long article? Highlight any passage and ask a follow-up question without leaving the page. The agentic features change daily habits the most. Comet can move through multiple websites on your behalf, fill in forms, track a purchase, and compile the results, and it connects to Gmail and Calendar for context. All of it runs on the free tier, which makes Comet the only fully free agentic browser available in 2026. ChatGPT Atlas needs a paid subscription for agent mode and runs only on macOS. Dia is macOS-only too, with no comparable agentic shopping or research workflow. Switching is low-friction. Comet supports Chrome extensions natively, so Grammarly, 1Password, and any other extension you depend on keep working. You can sign in with an existing Perplexity account or create a free one, and your search history and preferences then sync across devices automatically. Two limitations deserve a mention before you switch. Comet's DevTools are less mature than Chrome's, which matters if you live in the console. And some of Comet's shopping automation ran into a legal dispute with Amazon in early 2026. A Ninth Circuit stay lets Comet keep operating while the appeal proceeds, but the specific Amazon automation features have been affected, so check their current status at download time if that use case matters to you. Comet is what happens when a company that truly thinks about how people gather information builds a browser from scratch. It isn't a browser with a chatbot attached. It's a research environment that also happens to load web pages. Who should actually switch The right answer depends on how you spend your time in a browser. Here's the honest breakdown: Researchers, writers, analysts, and other knowledge workers who read multiple sources every day and suffer from tab overload: Comet is a real upgrade. Run it on your normal workflow for three days. The synthesis feature alone changes how you handle multi-source research. Former Arc users who want Arc's organizational workflow without the nostalgia: choose Dia if you have an Apple Silicon Mac. Choose Zen Browser (free, open-source, Firefox-based) if you want cross-platform support and the closest recreation of Arc's interface. Developers who rely on Chrome DevTools constantly: stay on Chrome. Gemini in Chrome offers AI sidebar features without a browser switch, and the developer-tooling gap between Chrome and every alternative is still real. Privacy-focused users who don't want to sign in to a search engine's browser: pick Brave with Leo (free, on-device AI conversations) or Firefox. Neither matches Comet's AI depth, but neither requires an account. Enterprise teams on Jira and Confluence: hold off until Dia gets its Windows release and enterprise features. Atlassian is clearly building toward deep Jira and Confluence integration, and that's the browser Atlassian is paid to care about. One point is worth stating plainly: Chrome's moat has shrunk. In January 2026, Chrome quietly added native vertical tabs, which had been one of the main arguments for moving to Arc or SigmaOS. Gemini in Chrome then brought a persistent AI sidebar with multi-tab comparison, saved prompts equivalent to Skills, and cross-app integration with Gmail and Calendar. The gap between Chrome and its challengers is narrowing from Chrome's side as well as theirs. The verdict The title asks whether any AI browser is worth switching to. The answer is yes, with a clear ranking and some honest caveats. Comet is worth the move if your browser is mainly a tool for research and reading. It's free, cross-platform, and compatible with Chrome extensions, and it comes from a company with $450M in ARR and a clear thesis about how information should work. Download it, use it for a week on your real workflow, and the decision will make itself. Dia deserves a try if you're on an Apple Silicon Mac and curious what an AI-native browser looks like with serious enterprise resources behind it. It isn't Arc, but nothing else offers Skills, and Atlassian's backing means it will keep improving in ways that no-revenue indie browsers can't match. Arc is worth keeping if you're already on it and your workflow is stable. Starting from scratch on it in 2026 isn't. SigmaOS is worth a look if you want Arc's UI philosophy and have accepted the funding risk. As your primary browser, it isn't worth $20 a month if you're choosing fresh in 2026. The browser market hasn't been this interesting, or this precarious, since Mozilla launched Firefox in 2004 and people first realized they didn't have to use Internet Explorer. Most of the new entrants won't survive the next two years. One or two will change what a browser means. Right now, Comet looks the likeliest of them all to be among that second group. Enjoyed this read? Be sure to follow me for more stories that make people feel things. You can also connect with me on LinkedIn or Telegram @kawsarlog for more updates on my work. I do freelance scraping projects, so if you need some data collected feel free to contact me on LinkedIn or email at hi@kawsarlog.com