← All journal entries

Local AI · Earlier experiment

Reading text from images with Ollama

I liked GLM-OCR when I tried it through Ollama. Looking at a simpler LM Studio setup then raised a different question: would the same model work in the tool I wanted to use?

One of my early local-AI experiments was trying GLM-OCR through Ollama. I was interested in reading text from images and documents, and my notes from the time record that I liked the result.

When I was looking at moving towards LM Studio and AnythingLLM, compatibility became part of the decision. My April setup notes flagged that GLM-OCR was not working in LM Studio at that point. They suggested another OCR model, or keeping Ollama specifically for that job.

This was a practical trial, rather than a scored OCR comparison. What stayed with me was the question: is a useful result worth keeping another tool around? The compatibility note belongs to that April experiment; software moves on.

April 2026 guide; 8 August 2026 retrospective capability profile

What came out of it

Tried GLM-OCR in Ollama and liked the result; the planned simpler stack raised a model-compatibility trade-off.