New Chinese Language Models

GLM 5.0 and MiniMax 2.5 have been released, two large language models (LLM) developed in China. Both models appear to be evolving from simple chats towards more complex, agent-style task execution workflows.

GLM 5.0 focuses on improving reasoning and code development capabilities. MiniMax 2.5, on the other hand, emphasizes the decomposition of tasks into simpler sub-tasks and the management of long-running executions.

Competition Evolves

It seems that the competition between language models is changing. It's no longer just about who can formulate the best answers, but about who is actually able to complete an entire workflow, automating complex tasks.