As of May (I/O) 2025, Google offers the Gemini app in three tiers. Many Gemini app features are available for free, while paying $19.99 per month for Google AI Pro (which includes 2 TB of storage) unlocks a lot more. Google AI Ultra at $249.99/month gives you everything else, including cutting edge capabilities.

Update 6/15: Refer to table

[Updated] Gemini models

All Gemini app users get “general access” to 2.5 Flash, which is currently the default model. Free users get “limited access” to Gemini 2.5 Pro (preview). Still in testing and not yet generally available (GA), Google says this model is for “Reasoning, math & code,” with Canvas particularly benefiting from it.

Google AI Pro subscribers get “expanded access” to 2.5 Pro (preview), which we now know translates to 100 queries per day. The company explains model limits as such:

Advertisement – scroll for more content

Gemini Apps has more prompt and chat limits for more advanced models. If you reach your capacity limit for a specific model for a given period, you can switch to a different model until that limit is reached or your capacity limit is refreshed. Gemini app users with Google AI Pro or Google AI Ultra have higher capacity limits for advanced models.

Google AI Ultra provides the “highest access” to 2.5 Pro (preview). A 2.5 Pro Deep Think mode will be available to Ultra subscribers “in the coming weeks,” while Agent Mode is another upcoming capability.

Context window

The context window on the free tier, which applies to all models, is 32,000 tokens or “around 50 pages of text.”

Google AI Pro and AI Ultra users get a long context window of 1 million tokens, which is equivalent to 1,500 pages of text or 30,000 lines of code.

File upload and analysis

Free users can upload documents and slides to get summaries, insights, and ask questions. Supported file formats include:

• Document files: DOC, DOCX, PDF, RTF, DOT, DOTX, HWP, HWPX
• Documents created in Google Docs
• Plain text files: TXT
• Presentation files: PPTX
• Presentations created with Google Slides

Uploading spreadsheets and other data files requires Google AI Pro and AI Ultra, with the ability to analyze and visualize (with charts) that information.

• Spreadsheet files: XLS, XLSX
• Spreadsheets created in Google Sheets
• Tabular data files: CSV, TSV

Similarly, the ability to upload code folders and repositories also requires a subscription. Google touts how you can get “insights from thousands of lines of code, make intelligent changes, debug errors, and optimize your code for peak performance.”

• Code files including C, CPP, PY, JAVA, PHP, SQL, and HTML

Reference past chats 

Free users have access to Saved info to specify chat preferences (like “I’m a vegetarian” or “Make responses concise”) for every conversation instead of having to add instructions to every prompt.

With Google AI Pro and AI Ultra, Gemini can look at past chats you’ve had to inform the current conversation. To trigger, “mention topics or timeframes of past chats” to trigger, while this capability can be used to summarize previous chats. The “Sources and related content” section will note when “Previous chats” have been used.

Deep Research

Gemini’s first agentic lets users ask a question and get a multi-point research plan that they can further customize. Once approved, Gemini will search the web, analyze what it found, and compose a report. As of I/O 2025, files and images can be uploaded to Deep Research, thus combining user and public knowledge for the final result.

• Free users: “Limited access” to Deep Research, which is is now powered by Gemini 2.5 Flash
• Google AI Pro: “Expanded access” using 2.5 Pro
• Google AI Ultra: “Highest access”

Audio Overviews

• Free users: Limited access
• Google AI Pro: Expanded access
• Google AI Ultra: Highest access

Image generation

All users have “general access” to image generation, which includes creating images with people. As of I/O 2025, the Gemini app is using Imagen 4 for higher quality, richer details, and better text/typography.

There’s also native image editing where you refine pictures (including generated ones and those you’ve uploaded) via text prompts.

[Updated] Video generation

• Free users: Not available
• Google AI Pro: Three 8-second 720p clips generated using Veo 3 Fast per day. Veo 2 afterwards.
• Google AI Ultra: Powered by Veo 3 with clips featuring sound (effects, noises, etc.)

Both videos use the same prompt: “Aerial shot of a grassy cliff onto a sandy beach where waves crash against the shore, a prominent sea stack rises from the ocean near the beach, bathed in the warm, golden light of either sunrise or sunset, capturing the dramatic elevation change and the serene beauty of the Pacific coastline.”

[New] Scheduled actions

Google is slowly rolling out “Scheduled actions,” with users able to have 10 active at any one time. You specify when (time/day/date or after an event) a predefined prompt should run. Afterwards, tap the mobile notification to see the response. 

Once live, you can edit, pause, and delete them:

• On the web: Settings & help > Scheduled actions
• On mobile: Profile menu > Settings > Scheduled actions

Other features

• Gems: To build custom versions of Gemini for specific tasks with pre-defined instructions
• Gemini Live