vLLM v0.25.0 Released: Model Runner V2 Now Default

Core summary

One sentence to understand this update

vLLM v0.25.0 is out, featuring 558 commits and making Model Runner V2 the default for all dense models, building upon previous quantized model support.

Impact & opportunity

What this could mean

Developers using vLLM will benefit from improved performance and efficiency with Model Runner V2 now being the standard for dense models, further enhancing quantized model inference.