Summarization

Why It Exists

Users are overwhelmed by long-form content and need quick access to key information without reading entire articles or documents.

How It Works

Summarization is Google's on-device AI text summarization feature powered by Gemini Nano. It reads long-form content, identifies key topics, extracts important sentences, and generates concise summaries. The system works with web articles, documents, emails, and chat conversations. All processing happens locally on supported Pixel devices using the Tensor NPU, ensuring privacy. Summarization was introduced as part of the Gemini Nano suite and is particularly useful for productivity and news consumption.

Everyday Use Cases

• Article summarization
• Document key points
• Email summarization
• Chat conversation summaries
• News digest
• Research paper abstracts
• Report generation

User Workflow

• User selects text or opens a document in a supported app
• Selects the Summarization option
• Gemini Nano analyzes the content on-device
• AI generates a concise summary highlighting key points
• Summary is displayed and can be saved or shared

AI Processing Flow

Text content is tokenized and analyzed by on-device Gemini Nano language models. The system identifies key entities, topics, and relationships across paragraphs. A neural extractive and abstractive summarization model generates the summary. The process uses the Tensor NPU for accelerated inference. All content stays on-device with no upload.

Inputs / Outputs

Inputs:

• Article text
• Document content
• Email threads
• Chat conversations
• Web page content
• Notes and memos

Outputs:

• Concise summary
• Key topic list
• Important sentence extraction
• Bullet point highlights
• Action item identification
• Shareable summary

Known Limitations

• Quality varies with content complexity
• May miss nuanced points in long documents
• Best results with English content
• Accuracy decreases with very short or very long texts
• Not available on non-Pixel devices

Unsupported Scenarios

• PDF files with scanned images
• Multilingual mixed-content documents
• Real-time conversation summarization
• Password-protected documents
• Code documentation

Performance Notes

Text analysis in under 1 second per 1000 words on Tensor G3; Summary generation in 2 to 3 seconds; NPU acceleration for language model; Battery consumption under 2 percent per summarization

Available On

Shows where this feature is available and how its AI processing works on each platform. Availability may vary by device.

Research Status

Confidence and verification reflect how complete documentation is. Fields may show Not assessed or Not yet verified while research is ongoing - this flags gaps, not product deficiencies.

Research Notes

Available on Pixel 8 and newer; Requires Google Tensor G3 or later; Introduced with Pixel 8 Pro 2023; Uses Gemini Nano 1 language model; Supports 5+ languages; Enhanced with Gemini Nano 2 on Pixel 9

Details

Capability Mapping

This feature maps to canonical capability family: Text Summarization.

Canonical capability: Text Summarization.

• Text Summarization - Condenses text, messages, documents, or conversations into a shorter summary.

Capability mapped; no device-level support assertions on file.