Frequently Asked Questions

Neon is an AI-powered browser that does far more than simply answer questions. It understands what you're working on and takes action on your behalf by opening pages, researching, filling in forms, and creating, with you in control at every step.

If you're just starting to explore the capabilities of AI, or even if you're already building complex workflows, Neon is ready to work with you. The more context it has about what you're doing, the more useful it becomes.

Neon runs on Opera’s AI engine, which is model‑agnostic, using different models depending on the task. Opera AI intelligently routes your task to the most appropriate model each time, and the model lineup is constantly updated so that you always have access to the best model for the job.

In Neon Chat you can also specify which model you'd like to use. Currently, Opera Neon gives you access to Gemini, Grok, GPT and GPT Pro, Claude Opus and Claude Sonnet, Deepseek, GLM-5 and Qwen3, along with creative models like Veo 3.1, Nano Banana 2 and Nano Banana Pro. With Neon Do, the agent picks the right model for the job. If you ask it to generate an image, it uses an image model; if you ask it to do research, it uses the best model for browsing and synthesis based on the context. Check our community on Discord for the latest updates.

What are the main features of Opera Neon?

Opera Neon’s main features are built around AI agents that help you chat, take action, create, and research directly in the browser.

Neon Chat is for everyday AI conversations and creative tasks, using your browser as context. It can see the pages you have open, so instead of copying and pasting content into a chat window, you just open the pages and ask Neon Chat to "compare these three articles," "find the contradictions between these two docs," or "summarize what's across these tabs." Neon Chat also supports creative features such as image generation, video generation, photo-to-video generation, and the creation of web apps, so you can use it not only to get answers, but also to create visual content and interactive projects.

Neon Do is focused on taking action in the browser. It can work with the page you are viewing, help with tab commands, support scheduled tasks, and connect with Google tools. It’s especially useful when you want Neon to help you complete tasks in the browser, rather than just answer questions.

Research Agents are specifically built to collect, analyze, and interpret data. Opera Neon includes three separate Research Agents for different needs.

• Deep Research is for detailed, in-depth research and structured reports.
• One-Minute Research is for quick research-style answers when you need a fast overview.
• Local Research uses your browser context, including Google Search results based on your location and preferences, and can include sources you are logged into, such as paid subscriptions.

Neon Cards are like skills. They instruct Neon how to handle a specific type of task without you having to explain it each time. They're organized into decks grouped by area of work, so you can find what fits your situation quickly. You can use cards from the library, or save your own. When you build a workflow that works well for you, you can turn it into a card and reuse it anytime with one click. Cards work across Neon Chat, Neon Do and all Research Agents.

MCP, or Model Context Protocol, is an open standard that lets different AI tools communicate directly with each other. Opera Neon acts as an MCP server, which means external AI tools that support MCP can connect to your live Neon browser session.

In practice this means the AI tools you choose to connect with can read the content of your open tabs, see what you're working on, take screenshots, and, when you allow it, can take actions inside the browser on your behalf. The key difference from other AI integrations is that Neon uses your real, logged-in browser session, not a sandboxed environment. So when an external AI connects through Neon, it's working with your actual accounts and live context.

Tools that currently work with Neon via MCP include Claude Code, Lovable, and n8n. You stay in control at all times. By default the connection is read-only, and actions like clicking or filling in forms require your explicit permission.

Tasks are a core concept in Neon. A task is a group of related tabs that share a common context, like when you’re planning a trip, going down a Wikipedia rabbit-hole, shopping, or opening new pages from an article you’re reading.

Within a task (a group of tabs), Neon can open and manage multiple tabs, knowing that they’re related. If you ask “What should I prioritize on this itinerary?” Neon knows it’s about your Lisbon task, your open hotel and flight pages, and other related tabs in the same group, and it can use all of those tabs as context for your question. You can create a task explicitly, or Neon can suggest it when it detects multi‑step work.

Your privacy and security is the first priority in Opera Neon. To ensure your privacy is respected, Neon is designed so that sensitive information such as your log-in credentials, passwords, credit card details, and cookies stay on your device as much as possible, and are never sent to our AI Engine. In order to function, Neon Do needs to understand the content of webpages relevant to the projects you assign it. Neon avoids using screenshots of your tabs as much as possible, and instead utilizes the structure of the webpages themselves (DOM tree and layout data). Only the webpage content is sent to Opera’s and OpenAI’s servers, and all such information is deleted from Opera’s servers after 365 days. Opera never trains AI models on your data. 

Neon Do runs locally within your browser and directly interacts with webpages. However, it uses cloud-based large language models (LLMs) to generate the plans and instructions it follows.

Neon Chat runs in the cloud. Research Agents also run in the cloud, with one exception: Local Research runs within your browser session, so it can use your local browsing context, including Google Search results based on your location and preferences.

Neon Do operates within your browser session, including pages you’re logged into and other session-based sites. However, when you sign into a site and enter your credentials, those credentials are not sent to any third-party servers (such as Opera’s). The same is true for payment details, which stay within your browser session and are never sent to Opera or any third‑party servers.

Data is retained on Opera’s servers for 365 days, allowing you to return to and continue chats and agentic workflows. After 365 days, dormant chats and workflows are deleted from our servers automatically. You can review and delete your Neon Chat and Neon Do activity through your chat window anytime. You can also delete your Opera account any time, which will also delete all Neon data associated with that account.

Opera Neon is a paid product and requires a subscription.

Yes, you need an Opera Account to use Opera Neon, as some browser functionality requires basic data from the account. However, an Opera Account is completely free and creating it only takes a minute. You can create a free account easily by visiting the Opera account main page.

Opera Neon is currently available for macOS and Windows. You can interact with AI Chat and use prompts in any language, and the user interface is currently available in approximately 40 languages.

Yes, Opera Neon will have a mobile version in the future.

Neon Do shows you the steps it plans to take. You can also observe what it’s doing in real time by watching the webpage(s) it's operating on. At any point, you can intervene to pause the process, make changes, or redirect its actions.

Opera doesn’t own or claim ownership of any output, including intellectual property rights, and can’t guarantee those rights either. You use Opera Neon and its AI Agentic features at your own risk. You’re responsible for anything that happens while using Opera Neon, including the content you upload and how you use any output from Opera Neon. You can read more in the Opera Neon Terms of Service.

No, Opera does not reserve any rights to your content, either input or output. When you submit content to Opera Neon, you consent for it to be used for the sole purpose of providing the service. Neon also has functionalities that allow you to publish content (e.g. Neon Cards). By using such features, you consent to the content being used for those services, including display and promotion.

Yes. Neon can learn and remember your personal preferences, such as shopping habits, travel preferences, favorite places, dietary needs, or other long-term context, and use them to make future responses more personalized and relevant.

You can teach Neon directly by saying things like “Remember that I prefer boutique hotels when I travel.” You can also let Neon save useful long-term preferences from your conversations automatically, or use the Save Memory button on a specific message. You can also view, edit, or remove saved memories in Settings, and ask Neon to forget something whenever you want. In addition, Neon lets you manually bring in prior context from other AI tools, such as ChatGPT, Claude, or Gemini, using provider-specific methods, so you don’t have to start from scratch when switching to Neon.

No. Opera doesn’t train models on user data. Opera’s AI engine orchestrates third‑party models, and Opera’s agreements with providers such as OpenAI and Google prohibit them from using Opera users’ data to train their models.

You can see Neon Do operate in real time. There are also buttons and visual cues, such as a subtle glow around active tasks and a side‑panel view showing the plan and next steps. You can pause, stop, or take over at any time. Neon Chat displays which tab(s) or task it’s referencing within the chat interface, so you always know what context it’s using.