v0.30.0 v6.4.2

Name: vllm-omni
Author: vllm-project

vllm-project/vllm-omniSep 25, 2026 → Sep 25, 20260 days between releases

Generating AI comparison

Analyzing what changed between v0.30.0 and v6.4.2. This page updates automatically — usually within a minute.

The Two Releases

v0.30.0

Sep 25, 2026· 26 days ago

A major update introducing a unified full-duplex serving framework, native cross-stage KV transfer, and interactive world-models, alongside expanded hardware support for Nvidia Blackwell and Ascend NPUs.

Full release notes

v6.4.2

Sep 25, 2026· 26 days ago

Optimizes the 'writing-plans' skill to generate leaner, faster plans by recording decisions (signatures, assertions) rather than full code transcripts, significantly improving speed and token efficiency.

Full release notes