Inlay

And now glm-5.2. This is wild artificialanalysis.ai/articles/glm...

I've been exploring open-source LLMs for agentic assistants since December, and I'm blown away by the latest batch of models (Kimi K2.7 Code and Minimax-M3). It's the first time I sense no practical difference, for my use, between them and the flagship models (Fable excluded) 1/2