Opera is taking a bet: turn the humble web browser into an active, working partner — and charge you for it. On Sept. 30, 2025, the Oslo-based company began inviting a limited group of early users to try Opera Neon, a so-called “agentic” browser that doesn’t just answer questions but can act on your behalf inside your own browser session. Access during this initial rollout comes with a subscription fee of about $19.90 a month and a waitlist for everyone else.

Opera pitches Neon as a browser built around a handful of AI primitives that map to real work. The most visible are Tasks — self-contained workspaces that group relevant tabs, chats and context for a single activity — and Cards, reusable prompt blocks you can combine into repeatable automations. The feature that’s likely to get headlines is Neon Do: an agent that can open and close tabs, crawl sites you’ve given it access to, fill in forms and compare or extract information across pages — and it does a lot of that locally in your browser, not out in an opaque cloud. That local-first framing is central to Opera’s pitch about privacy and control.

Opera’s demos show Neon doing things like summarizing articles, pulling data from a video you watched last week, or combining “pull details” + “comparison table” Cards to produce neat side-by-side product comparisons. The company calls the result an “agentic” experience — AI that takes steps for you rather than only recommending them.

Price and availability: early access, paid model

Unlike most mainstream browsers, Neon is a premium product. Opera says it will invite a small number of users today and expand invites over time; everyone else can join the waitlist. Industry reporting pegs the subscription at roughly $19.90–$19.99 per month, a price Opera appears to be using to position Neon as a tool for “power users” and professionals who use AI heavily in their workflows.

That pricing is notable because many competing AI-baked browsing tools are free (or bundled into larger services). Opera’s move is an explicit bet that a segment of users will pay for a more cohesive, privacy-minded agentic experience and faster rollout of advanced features.

Where Neon sits in a suddenly crowded field

Neon didn’t arrive in a vacuum. The past year has seen several companies try to make browsers smarter in different ways:

• Perplexity’s Comet is already shipping as an AI-first browser aimed at doing research and workflows inside a browser shell.
• OpenAI’s ChatGPT Agent moved into the “do things for you” lane earlier this summer, giving agents limited ability to interact with the web and local tools.
• The Browser Company (makers of Arc/Dia) has become a hot commodity; Atlassian announced a deal to acquire it in September, a sign of how strategically valuable this space looks to established enterprise players.
• Google is also folding its Gemini models into Chrome, adding context-aware assistance and automated workflows to the world’s dominant browser.

Put simply, everyone from scrappy startups to Google and Atlassian is pushing at the border of “browsers that act,” and Opera is the latest mainstream browser maker to make a hard product play rather than merely bolting on an assistant.

The promise — and the obvious questions

Neon’s promise is seductive: tidy workspaces, a set of composable prompts you can reuse, and an assistant that can operate directly in the environment you already use (your browser session, logged-in accounts and all). Opera emphasizes that Neon’s agent runs visibly and locally, and that users stay in control — you can pause or take over at any moment. That’s a practical response to the big trust problem around agentic AI: invisible services doing things on your accounts.

But there are a bunch of unanswered and practical questions that will decide whether people actually pay:

• Is $20 a month worth it? Chrome and other browsers are getting better AI features for free; a paid browser needs to prove it’s materially faster, safer, or more private for the kinds of workflows that make people productive. Early critics and reviewers are already asking whether Neon’s unique features justify the price.
• How well will Neon Do work at scale? Demos are useful, but they rarely match messy, real-world sites and authentication flows. Neon’s effectiveness will be measured by the hairier tasks — automating purchases, dealing with paywalls, or interacting with flaky web forms. TechCrunch cautioned that demos don’t always translate into day-to-day reliability.
• Privacy and safety tradeoffs. Opera sells Neon as privacy-sensitive — the agent runs locally and doesn’t require cloud password handoffs — but real users will want transparency about telemetry, how Cards are shared, and whether any on-device data might be used for model training. Opera’s documentation addresses some of this, but the fine print matters.

Opera Neon is an interesting, well-scaffolded attempt to make the browser itself the platform for agentic AI. The Tasks/Cards/Do model is a tidy mental model for non-technical users, and Opera’s emphasis on local, visible action is a welcome counterpoint to fully cloud-based agents.

That said, the price and crowded competitive landscape are Opera’s two biggest hurdles. If Chrome, Comet, Dia (now under Atlassian) and ChatGPT-style agents offer similar convenience for free — or if Neon’s workflow advantages aren’t dramatically better in practice — convincing $20-a-month shoppers will be an uphill run.

If you’re professionally dependent on rapid, repeatable web workflows and you care about the local-first privacy framing, Neon is worth watching (and maybe joining the waitlist). If you’re a casual user or happy with Chrome/Edge/Safari plus a handful of AI tools, Opera still has to make a very practical case for this subscription.