Model guides

How to use Gemini 3.1 Pro online with photos and files

Google describes Gemini 3.1 Pro as a model that reads text, images, audio, video and PDFs. Here is how to use it on your own files and what to check.

On this page
  1. What Google documents about Gemini 3.1 Pro
  2. Gemini app, Google AI Pro or a model picker
  3. Fictional example: a whiteboard photo and a project brief
  4. Check the transcription, references and gaps
  5. Get a clearer read from photos and screenshots
  6. What Google's Gemini app offers that this does not
  7. Use Gemini 3.1 Pro in Chat AI App
  8. Common questions
  9. Sources

To use Gemini 3.1 Pro online, open Google's Gemini app or choose Gemini 3.1 Pro in Chat AI App's model picker. Google said in February 2026 that 3.1 Pro was rolling out globally in the Gemini app, with higher limits on Google AI Pro and Ultra. The second route is a separate subscription that puts Gemini next to GPT, Claude and other model families.

Either way, the useful work is the same. Attach the image or document, ask a question with a checkable answer, and verify the result against the source. This guide covers what Google documents, a fictional photo-plus-PDF example and what Google's own app includes that a model picker does not.

What Google documents about Gemini 3.1 Pro

Google's Gemini 3.1 Pro model page lists it as a preview model. It accepts text, images, video, audio and PDF files as input and returns text. Google presents it as a refinement of the Gemini 3 Pro series, with improved reasoning and more grounded, consistent answers. The page also lists API features such as Google Search grounding, code execution and URL context.

For files, Google's document understanding guide says Gemini models read PDFs with native vision. That covers text, images, charts and tables inside the document. Its image understanding guide covers captioning, classification and visual question answering. It also shows prompts that combine several images with text.

These pages describe Google's models and API. They do not set the file limits, tools or behavior of any other app. Treat them as a guide to what the model can do, then test it on your own file.

Gemini app, Google AI Pro or a model picker

Google's limits and features page sets out what each Gemini app plan adds. Without a plan, you get standard limits and a 32k-token context window. Google AI Pro raises limits to four times the standard and the context window to 1 million tokens. The table puts those options next to a multi-model app.

OptionWhat it offersWorth considering when
Gemini app, no plan3.1 Pro, per Google's February 2026 release note, with standard limits and a 32k-token context windowYou ask occasional questions about short files.
Google AI ProHigher 3.1 Pro limits, a 1 million-token context window and expanded access to Gemini app features and modelsYou rely on Google's own app features or work with very long files in one chat.
Chat AI AppGemini 3.1 Pro and Gemini 3.7 Flash in the picker, beside GPT, Claude and other models, plus uploads, web search and downloadsYou want several model families and your files in one conversation.

Fictional example: a whiteboard photo and a project brief

Fictional example: Ana runs events for Riverside Arts Trust, an invented charity. After a planning meeting, she has a phone photo of the whiteboard and the 12-page event brief as a PDF. The whiteboard says load-in is Thursday, Sam owns ticketing and the raffle is dropped. The brief says load-in is Friday and lists the raffle on page 9.

She picks Gemini 3.1 Pro, attaches both files and sends this prompt:

The image is a photo of today's planning whiteboard. The PDF is the event brief, version 3. First, transcribe every whiteboard item as a list. Mark any word you cannot read as [unclear] and do not guess. Then compare each item with the brief and label it "matches", "conflicts" or "not in brief". Give the brief's page number for every match or conflict. Do not resolve conflicts. End with a table: whiteboard item, brief page, status and question for the team.

A correct answer flags two conflicts: the load-in day and the raffle. Sam's ticketing role is labeled "not in brief" unless the brief names a different owner. Smudged words appear as [unclear]. If the table says load-in is settled for Friday, the model chose between sources, which the prompt told it not to do.

Ana then asks for the final table as an Excel file and shares it with her team before the next meeting.

Check the transcription, references and gaps

  • Transcription: Compare the list with the photo. Handwriting, glare and cropped edges cause misreads.
  • Conflicts: Check that both sides of each conflict are shown and that none was quietly resolved.
  • Coverage: Count the whiteboard items yourself. A missing line is easy to overlook in a neat table.
  • Unclear items: Fill in [unclear] words from your own notes or a colleague, not from the model.
  • Arrows and grouping: Check that items joined by arrows or boxes on the whiteboard stay together in the list.

Get a clearer read from photos and screenshots

Most misreads start with the image, not the prompt. A few minutes of preparation saves a round of corrections.

  • Shoot straight on: Hold the phone parallel to the board or page, and avoid glare from windows and lights.
  • Crop to the content: Cut away walls, desks and other boards so only what matters is in frame.
  • One image per question: If you send several photos, number them in the prompt and say which is newest.
  • Screenshots beat photos of screens: A screenshot keeps text sharp; a photo of a monitor adds blur and moiré.
  • Keep the uncertainty rule: Repeat the [unclear] instruction in every prompt so guesses stay visible.

What Google's Gemini app offers that this does not

Choosing Gemini 3.1 Pro in Chat AI App's model picker works inside that app's workspace, not Google's app. Google's limits page lists Gemini app features such as Deep Research, Canvas, connected apps, audio overviews and video generation. Those Google features do not come with the model. Where the multi-model app has a tool for a similar job, such as web search, it is its own tool.

Google's lineup is also moving. Its release notes say Gemini 4 Argon began rolling out to Google AI Pro and Ultra subscribers on October 13, 2026. Gemini 4 Argon is not one of the picker labels in the table above as of this update. If you need Argon or Google's native features, evaluate Google's own plans.

The multi-model route fits a different need: Gemini next to GPT, Claude and other models in one conversation. It adds uploads, web search with sources and downloadable files. Compare the two on one real task before you decide. Our Gemini document guide has another workflow to try.

Use Gemini 3.1 Pro in Chat AI App

  1. Go to AI Chat in your browser, on a laptop or a phone.
  2. Attach the photo with + → Photos and the PDF with + → Files first. One message can carry several files, including Word, Excel, CSV and PowerPoint files.
  3. Then choose Gemini 3.1 Pro from the model picker. Gemini 3.7 Flash sits in the same list.
  4. Wait for the attachments to finish processing, then send your prompt.
  5. Check the answer with the list above. For a second read, switch to GPT-6 or Claude Opus 5.5 without starting the conversation over.
  6. Export the final table to Excel, or download the answer as a Word or PDF file. Then open the file and check its formatting.

Chat AI App is independent of Google, so no Google AI plan or Gemini app feature comes with it. You can pay $29.95 a month, $66.50 a quarter or $179.00 a year, and cancel anytime. Full terms are on the pricing page.

Common questions

Can I use Gemini 3.1 Pro without Google AI Pro?
Yes. Google's release notes say 3.1 Pro rolled out globally in the Gemini app, with higher limits on Google AI Pro and Ultra. In Chat AI App, Gemini 3.1 Pro appears in the model picker under a Chat AI App plan, with no Google AI subscription required.
Is Gemini 3.1 Pro Google's newest model?
No. Google's release notes say Gemini 4 Argon began rolling out to Google AI Pro and Ultra subscribers on October 13, 2026. Google's API page still lists Gemini 3.1 Pro as a preview model.
Can Gemini 3.1 Pro read a photo of a page or a scanned PDF?
Google documents image and PDF input for the model, and Chat AI App accepts images and scanned PDFs. Check the transcription against the original, and ask the model to mark unreadable words instead of guessing.
Does Chat AI App include Deep Research or Gemini's connected apps?
No. Those are features of Google's Gemini app. Chat AI App offers its own web search with sources, file uploads and downloads. It can also read Google Docs, Sheets or Slides links that are shared publicly.

Sources and further reading