vLLM v0.25.0 released, with Model Runner V2 as default for dense models.

Core summary

One sentence to understand this update

vLLM has released version 0.25.0, featuring 558 commits from 232 contributors and making Model Runner V2 the default for all dense models.

Impact & opportunity

What this could mean

Developers using vLLM can expect improved performance and efficiency for dense models with the new Model Runner V2, enhancing their LLM serving infrastructure.