apple/LensVLM-9B · Hugging Face

LensVLM is a 9B Vision Language Model (VLM) designed to process compressed images of text. It selectively expands only the relevant pages to their uncompressed form using learned tools. The model, available on Hugging Face as bartowski/LensVLM-9B-GGUF, has its accompanying source code distributed separately under the Apple Sample Code License.

This 9B Vision Language Model from Apple is the first to use selective expansion of compressed text images, unlike prior models that process all data uniformly.