Gemini API URL Context Tool Expands Support for PDFs and Images

AUG. 18, 2025 — Google has announced the general availability of the URL context tool for the Gemini API, enabling developers to ground prompts using web content, PDFs, and images. This tool eliminates the need for manual uploads, offering a more versatile and scalable solution for generative AI applications.

Key Features

• PDF Support: Developers can now link directly to PDFs, allowing Gemini to extract text, understand tables, and analyze document structure.
• Image Support: The tool processes images (PNG, JPEG, BMP, WebP), leveraging Gemini's multimodal capabilities to interpret visual data like charts and diagrams.
• Web and Data Files: Continued support for HTML, JSON, XML, CSV, and other text-based formats.

Production-Ready Scaling

The tool is now optimized for production use with updated rate limits and pricing. Costs are based on the specific Gemini model and the standard token pricing.

Use Cases

• Personalized Customer Interactions: AI chat agents can reference customer websites for accurate responses.
• Document Comparison: Analyze multiple reports or articles to identify trends.
• Content Synthesis: Combine information from multiple URLs to generate summaries or reports.
• Code Analysis: Point to GitHub repos or docs for technical explanations.

Real-World Examples

• Gemini CLI: Uses URL context for web-fetch commands, enabling tasks like summarization and translation.
• Gladly.ai: Enhances customer service by grounding AI agents with up-to-date website content.

Getting Started

Developers can explore the URL context documentation or try the demo app. A Python example is provided:

python from google import genai from google.genai.types import Tool, GenerateContentConfig client = genai.Client()

model_id = "gemini-2.5-flash" tools = [ {"url_context": {}}, ]

response = client.models.generate_content( model=model_id, contents="What are the top 3 recent announcements from the Gemini API according to https://ai.google.dev/gemini-api/docs/changelog", config=GenerateContentConfig( tools=tools, ) ) for each in response.candidates[0].content.parts: print(each.text)

Google encourages developers to experiment with the tool and share their creations.