Comment by walrus01

Take a look at Thomson 1.0-small, which is a variant of qwen 3.6 35b post trained by Thomson Reuters for text analysis. It classifies text content very well.