Opera Neon Agentic Browser (Standard Plan)

Opera Neon is the company's attempt to turn the browser from a window you look through into something that does the clicking for you. Rather than bolting a chat sidebar onto an existing browser, Opera has built the product around a set of distinct agents — Chat for conversation, Make for generating things, Do for carrying out actions on your behalf, and, since October 2025, the Opera Deep Research Agent (ODRA) for longer multi-step investigations. The pitch is that these agents read the page you're on and carry context between tabs, which is the part that separates an agentic browser from a bookmarked AI tab.

The audience here is fairly specific: Windows and Mac users who already spend a meaningful part of their day on research, comparison shopping, form-filling or content drafting, and who are willing to pay $19.90 a month for the Standard plan to unlock the interesting half of the product. Free users get the browser itself with ad blocking and VPN, the Opera Browser CLI, an MCP connector for wiring in third-party agents, and basic Chat — a genuinely useful free tier, but one that deliberately withholds Do, Make and deep research. If you mainly want a fast, private browser, the free download is the whole story. If you want the agents, this is a subscription decision, not a browser decision.

What impresses me most on paper is the architecture rather than the feature list. Opera's AI Engine being both model- and tool-agnostic — able to draw on different LLMs such as Gemini and GPT-class models — is the right structural choice in a market where the best model changes every few months, and the full model choice granted on the Standard plan turns that into a user-facing benefit rather than a backend detail. Splitting a research problem server-side across parallel researchers is also a sensible way to make deep research feel less like watching a progress bar. Shipping an MCP connector and a browser CLI is the unusually confident move: Opera is inviting you to plug in Claude or ChatGPT sessions alongside its own agents instead of locking you in.

The caveats are substantial, though. The performance case rests on Opera's own framing of DeepResearch Bench results, which is a vendor claim about a vendor's product and not something I can treat as settled. The roadmap talk — many more agents and a new orchestration layer internally called 'Symphony', which Opera's Krystian Kolondra says will fundamentally change multi-agent interaction — is an admission that the current interface for juggling four agents isn't the final answer. And at $19.90 a month, Neon is priced against general-purpose AI subscriptions that many readers already pay for, while the information available says little about reliability, speed, or how often the Do agent actually completes a task unsupervised. That last point is the whole product, and I can't verify it from specs.

I'd call this a promising, well-architected agentic browser with a transparent price and a more open posture than most of its rivals, held back by a thin evidence base and a UI the maker is already rebuilding. Worth a trial on the free tier; treat the subscription as an experiment rather than a replacement for your existing AI tooling. One honest caveat on the rating: this assessment is built on specs, pricing and company claims rather than extended hands-on testing, and several details — platform coverage, tier contents and the release timeline — come from lower-confidence secondary sources, so treat the score as provisional.