Anthropic Introduces Claude 3.5 Sonnet with Visual PDF Analysis for Images, Charts, and Graphs under 100 Pages

Claude has introduced a new feature preview that can read all kinds of visuals inside a PDF that is less than 100 pages. This now makes it easy to upload a document, retrieve the complete context, and digest information from PDFs, especially research papers and technical documents that contain charts and graphs, among other images and visuals. 

Visual PDFs are an experimental feature in the Feature Previews available on Claude 3.5 Sonnet. 

Claude can now view images within a PDF, in addition to text.

This helps Claude 3.5 Sonnet more accurately understand complex documents, such as those laden with charts or graphics.

Enable the feature preview: https://t.co/bJ8BjBT6zG. pic.twitter.com/VNSf547ptT

— Anthropic (@AnthropicAI) November 1, 2024

The good news doesn’t end there. The company has also increased the document size limit from 10MB to 30 MB. A user on X was quick to point out the change, and Claude now lets you upload a maximum of five images or documents, with a size limit of 30 MB each. 

“Up until today, when you attached a PDF in Claude.AI, we would use a text extraction service to grab the text and send that to Claude in the prompt. Now, Claude can actually see the PDF visually alongside the text”, said Alex Albert, Head of Claude Relations at Anthropic, in a post on X.

You can access the feature from the pop-up banner on the home page. Once you select the Visual PDFs in the Feature Preview tab and turn it on, it will be made available for future conversations. Moreover, Anthropic has also announced that it supports adding PDFs as input in an API request. 

Anthropic Is on a Roll

Only a few days before, Anthropic released Computer Use which caused quite a storm in the AI ecosystem. Recently, they also announced a partnership with GitHub, which included Claude 3.5 Sonnet inside GitHub Coilot. 

A few days ago, it was also announced that Claude can now execute and run JavaScript code. It’s called the Analysis Tool, and it can also generate data visualisations after it writes and executes the code. Apart from Visual PDFs and Analysis Tool, Claude also provides a feature called LaTex rendering to generate mathematical equations from a user’s input.

At this point, it is well established that Claude’s 3.5 Sonnet is the best AI model for running code. OpenAI’s latest GPT o1 isn’t there yet, and even with its latest offering, Canvas, it still fails to keep up with Claude’s abilities.