Google Gemini - AI Input & Output

We analyzed 36 examples of how Google Gemini handles ai input & output across their product, spanning 12 different approaches: entry point, input composition, model selector, attachment handling, voice input, input modifier, context indicator, citation display, loading state, message layout, split view, text formatting.

Entry Points

Google Gemini — gemini.google.com home
Gemini landing page with personalized greeting, input field with placeholder, starter prompt chips, attachment button, Tools button, model selector (Fast), and microphone icon

Attachment Handling

Google Gemini — gemini.google.com home
Attachment menu expanded from the + button showing three upload options; voice input error message displayed in the input field

Google Gemini — gemini.google.com home
Input composition with a PDF file attached and prompt text typed; file appears as a thumbnail card above the input text

Google Gemini — gemini.google.com home
Input composition with an image attached and prompt text typed; image appears as a thumbnail above the input text

Input Modifiers

Google Gemini — gemini.google.com home
Tools menu expanded showing available tool/mode options for configuring the AI's behavior

Model Selectors

Google Gemini — gemini.google.com home
Model selector dropdown expanded showing Gemini 3 model variants with descriptions and New badges

Voice Input

Input Composition

Google Gemini — gemini.google.com home
Input composition with a general knowledge query typed, submit button visible, no tools activated

Message Layout

Google Gemini — chat-conversation
User asked 'What happened in tech news today?' with Deep Research tool active. Gemini produced a research plan with steps (Research Websites, Analyze Results, Create Report) and action buttons to proceed or edit.

Loading States

Google Gemini — chat-conversation
Loading state immediately after submitting 'What is design systems?' query. Animated sparkle/thinking indicator shown where the AI response will appear.

Split View

Google Gemini — chat-conversation-split-view
Canvas mode active. User asked for a React modal component. Split-view with AI thinking/response on left and code editor canvas on right showing React code with line numbers.

Google Gemini — chat-conversation-split-view
Canvas mode completed. Full AI response visible on left with feature list, file card, and 'Try again without Canvas' link. Code canvas on right unchanged.

Google Gemini — chat-conversation-split-view
Canvas mode with Preview tab active. Right panel renders a live preview of the React modal component. Left panel shows same completed response.

Text Formatting

Google Gemini — chat-conversation
AI response streaming for 'What is design systems?' showing formatted text with bold terms, numbered sections, bullet points, and inline citation link icons.

Google Gemini — chat-conversation
AI response (non-Deep Research) to 'What happened in tech news today?' showing formatted text with emoji section headers, bold text, bullet lists, and inline citation links with hover tooltip.

Google Gemini — chat-conversation
AI response to 'What's in this image?' showing image analysis with the uploaded image displayed, an 'Analysis' collapsible section header, and formatted breakdown with bold labels and nested bullet lists.

Citation Display

Google Gemini — chat-conversation
Completed AI response for 'What is design systems?' showing bottom portion with data table, examples list, follow-up question, Sources button, action bar, and Sources side panel open.

Google Gemini — chat-conversation
Completed AI response summarizing an uploaded PDF document. Shows formatted text with citations from the document, a Sources button, action bar, follow-up question, and a Files side panel open on right.

Google Gemini - AI Input & Output

We analyzed 36 examples of how Google Gemini handles ai input & output across their product, spanning 12 different approaches: entry point, input composition, model selector, attachment handling, voice input, input modifier, context indicator, citation display, loading state, message layout, split view, text formatting.