The open source LLMs are working fine for image recognition. We have used the technology internally. For example, someone took a selfie in front of a mirror and

The open source LLMs are working fine for image recognition. We have used the technology internally. For example, someone took a selfie in front of a mirror and the description as a result of the prompt also recognized that there was a mirror image there.