In the previous few weeks, Anthropic has launched some thrilling beta options which have largely gone below the radar. Considered one of these was its new token-counting API. I’ve already written an article on this, which you’ll learn by clicking the hyperlink beneath.
The opposite thrilling characteristic, and the topic of this text, is that Claude 3.5 can now course of PDFs and perceive each textual content and visible content material inside PDF paperwork.
PDF Capabilities
Claude works with any normal PDF file, permitting you to inquire about textual content, photos, charts, and tables inside your paperwork. Listed below are some widespread use circumstances:
- Analyzing monetary studies, deciphering charts and tables
- Extracting key info from authorized paperwork
- Helping with doc translations
- Changing doc content material into structured codecs