Opera Neon is now available and aims to be your paid AI browser

Opera has quietly shifted one of its most experimental projects out of the lab and into people’s app folders. What started this fall as a “Founders” experiment is now a paid product: Opera has removed the Neon waitlist and opened the browser to anyone willing to pay $19.90 a month. The company pitches Neon less as a standard web viewer and more as a pocket-sized team of assistants — software that reads pages for you, checks other sources, and can even carry out routine actions like filling forms or completing checkouts.

That’s not just marketing spin. Neon was built from the ground up to let its AI actually manipulate the interface rather than live in a tiny sidebar. Instead of a single chatbot running on top of an ordinary browser tab, Neon’s agents can open and close tabs, hop between sites, scrape information and then act on it — the “Do” and “Make” agents are explicitly designed to perform sequences of tasks you’d otherwise do by hand. Opera describes the approach as a way to reduce the “mental overhead” of web work: hand off the mechanical stuff and keep the human part — judgment, nuance, verification.

That framing helps explain the price. Opera isn’t pretending Neon is a free, mass-market browser; it’s being sold as a subscription productivity tool, pitched at power users and professionals who might value hours reclaimed over an extra $20 a month. Early reviews and hands-on guides note that Neon is feature-rich and fast, but also that its value depends on whether those agentic features actually reduce friction rather than add new complexity to a user’s workflow. For someone who mostly needs a fast, standards-compliant browser, Neon’s promise may feel like gilding the lily. For a freelancer, researcher, or small team that spends a lot of time compiling information and doing repetitive web tasks, though, a reliable assistant that can orchestrate the web could be worth the monthly fee.

Related /

Part of Neon’s appeal to tinkerers is the way Opera bundles multiple models under one roof. Rather than forcing users to pick a single provider, Neon offers access to a roster — Google’s Gemini 3 Pro, OpenAI’s GPT-5.1, Google’s Veo 3.1 for video, and a model Opera calls Nano Banana Pro, among them — and layers its own agentic logic on top. The result is essentially a model buffet: pick the engine that fits the job and let the browser handle the plumbing. Opera argues this spares users the juggling of multiple subscriptions and the awkward UX of switching tools mid-task.

A new piece of that stack is ODRA, the Opera Deep Research Agent, which Opera designed for longer, structured investigations. ODRA’s headline trick is a “1-minute research” mode that runs parallelized searches across sources and returns a synthesized, citation-backed report. The pitch is blunt and pragmatic: what many knowledge workers do manually — skim a handful of articles, cross-reference claims, assemble links — should be compressible into a single, auditable output the browser produces for you. Opera’s materials emphasize clear sourcing rather than the cozy hallucinations that have plagued earlier AI summarizers. That’s a small but crucial difference if you’re going to pay for output you plan to reuse in work or reporting.

Neon isn’t arriving in a vacuum. The browser-as-agent idea is fast becoming an arms race: Perplexity’s Comet browser, The Browser Company’s Dia, and others are chasing adjacent visions of how much of the web a machine should be allowed to read and act on. Where some competitors prioritize an answer-first search experience or novel interface metaphors, Neon’s differentiator is a mix of deeper automation and the breadth of model access it offers. That’s a practical bet: if the selling point is “it gets things done for you,” then depth of automation matters as much as conversational polish.

That practical bet comes with practical questions. Agentic browsing raises the usual trio of concerns: accuracy, provenance and control. Opera’s ODRA aims to surface citations to address provenance, but users will still need to judge sources — a browser can assemble a dossier faster than a human, but it can’t replace the judgment calls about which citations deserve the most weight. Then there’s the security and privacy angle: Neon installs as a separate app and comes with Opera’s usual array of built-in tools, but handing a tool control over form filling and checkouts raises questions about credential handling, autofill protections, and where those action logs live. Opera’s documentation and press materials address some of these topics, but anyone planning to put sensitive workflows on autopilot should examine the settings and safeguards closely.

Getting Neon today is straightforward: with the waitlist removed, anyone can download the separate Opera Neon app from operaneon.com and subscribe. It won’t displace your day-to-day browser unless you make it the default, which is a sensible separation: use Neon for tasks that benefit from agentic help and keep your primary browser for casual surfing. Opera is also positioning Neon as a community-driven product, rolling regular updates and working hand-in-hand with its Founders community during the early weeks of public access.

The larger question is cultural as much as technical: are users ready to treat a browser like a paid, semi-autonomous assistant? Opera’s move tests a wedge between “browsing” and “delegating.” If ODRA and Neon’s other agents can consistently produce verifiable, time-saving outputs, the subscription will look like paying for a virtual research assistant that lives in your address bar. If they can’t, the product risks becoming an expensive novelty — a clever demo that doesn’t meaningfully alter day-to-day work. Either way, Neon’s public launch marks a clear moment in the evolution of the browser: the window to the web is being remade into a place where software not only shows you pages but can also act on them.