Mistral AI's Medium 3.5 model caused disappointment. The 128B parameter model is competitive in benchmarks but expensive. Chinese competitors GLM, Qwen are superior and free. User reactions are mix...
