Gemini in Chrome Select from Screen Lets Users Target Specific Page Content
Gemini in Chrome can now focus on exactly what a user points to, a chart, a comparison table, a diagram, rather than processing an entire page and guessing what's relevant. Google updated its Gemini in Chrome support pages this week to document "Select from screen," a new desktop tool that lets users draw a bounding box around on-page content before submitting a prompt. The feature is rolling out now in Chrome 149 to signed-in users who aren't browsing in Incognito, Android Authority reported today.
The workflow resembles Circle to Search: select something on screen, ask a follow-up, stay in context without starting over. Google said in February that Circle to Search handles billions of queries per month, per the Google Search blog. The feature had already reached more than 300 million Android devices by last summer, according to a separate Google Search post. "Select from screen" doesn't transplant Circle to Search to desktop; it adapts the same core mechanic, select a region, ask a question, stay on the page, to Chrome's Gemini side panel.
What follows covers what the feature actually does, when it's better than standard prompting, what remains unconfirmed, and who can use it today.
How Gemini in Chrome Select from screen works
The workflow is short. Open the Gemini side panel via "Ask Gemini," tap the plus icon for the Add menu, select "Select from screen," and draw a box around the content you want Gemini to see. Users can highlight multiple regions at once before sending their prompt, according to Android Authority, which means comparing two separate elements on a page doesn't require two separate conversations.
That multi-region selection is the feature's clearest differentiator. Gemini in Chrome already handles full-page summarization and can compare information across tabs from its persistent side panel, capabilities documented in Google's Chrome blog from two weeks ago and introduced with the January 2026 Gemini 3 update. "Select from screen" narrows the context deliberately. Rather than trusting Gemini to identify what's relevant across an entire page, users anchor the question to a specific element.
The practical difference shows up in scenarios where full-page context is noise, not signal. A data table buried midway through a long report. A product comparison grid on a retailer page. Two separate spec blocks a user wants to evaluate side by side. In each case, drawing a box around the relevant content is faster and more precise than describing it in text, and it eliminates the risk that Gemini anchors its answer to something else entirely on the page.
One genuine gap worth flagging: Google's support documentation, as reported by Android Authority, does not specify which content types the tool supports. Whether it works on images, embedded media, or only standard text and inline graphics is not confirmed. Keep expectations calibrated to what's been verified.
Gemini in Chrome's growing input toolkit
"Select from screen" arrives as Gemini in Chrome has been adding input methods steadily. The January 2026 update introduced the persistent side panel and auto browse for multi-step tasks, per Google's Chrome blog. A June expansion added the ability to schedule meetings via Calendar, check Maps, draft emails in Gmail, and ask questions about YouTube videos, all without leaving the current page, per the same blog. There's also Nano Banana 2 support, which lets users transform on-page images using a text prompt in the side panel.
"Select from screen" fits into this arc as a precision input layer. Where other features extend what Gemini can do after receiving a question, this one changes what the user provides before asking it. That's a different kind of upgrade: less about capabilities, more about accuracy of context.
The comparison to Circle to Search is useful here as a mental model. Circle to Search's multi-object update, which let users circle multiple items in an image simultaneously, arrived in February. "Select from screen" brings a similar logic to desktop browsing, where a mouse and keyboard make precise region selection more natural than a finger gesture on a phone screen. The mechanics fit the surface.
Who can use it and what the limits actually mean
Desktop access requires Chrome 149 or later, Gemini in Chrome enabled, and a signed-in Google account. The feature does not work in Incognito mode, per Android Authority.
Whether "Select from screen" is available to all signed-in users or tied to a paid subscription is not confirmed. Google has not specified a tier for this feature. For reference, other Gemini in Chrome capabilities do carry subscription gates: auto browse requires AI Pro or Ultra, and Personal Intelligence requires AI Plus or above, according to Google's Chrome blog from two weeks ago. Whether "Select from screen" follows the same pattern is an open question worth checking before assuming access.
The Incognito restriction deserves a plain-language read. The more Gemini's usefulness in Chrome depends on reading on-screen content, the less of that experience is available outside a signed-in session. Google states the feature is "built with security in mind" and that its models are trained to recognize prompt injection attempts, per the Chrome blog. Those are vendor assurances, not independent assessments, and worth keeping in that category.
One clarification for anyone who's seen Android-specific requirements elsewhere: the specifications tied to the Gemini in Chrome mobile rollout, Android 12 or higher, at least 4GB of RAM, English-US language setting, apply to the Android release announced in May. They have no bearing on desktop access to "Select from screen."
What to watch as the rollout progresses
For anyone already on Chrome 149: open the Gemini side panel, use the Add menu to find "Select from screen," draw around the content you care about, multiple regions if needed, and ask your question. For dense or visual pages where typing a description would be more work than it's worth, the feature appears well-suited to the task based on how Android Authority describes the documented workflow.
Two open questions will determine how broadly the feature lands. First, whether it requires a paid subscription or works for any signed-in Chrome user. Second, which content types it actually supports, since Google hasn't confirmed whether the selection tool handles images and embedded media or is limited to text and inline graphics. Both should become clearer as the rollout scales.
The Circle to Search trajectory is worth keeping in mind. That feature started as a simple tap-and-select tool, then expanded to multi-object image search, AI Mode follow-ups, and in-game assistance as it reached more than 300 million Android devices, per Google. "Select from screen" is at step one on desktop. The underlying interaction, point at something specific, ask a question, stay on the page, has already been used at scale on Android. Chrome's desktop version now has room to grow into it.
Comments
Be the first, drop a comment!