# vLLM 0.28 adds KV offloading and serving changes

_Published Saturday, August 29, 2026 at 6:06 PM EDT · Infrastructure · Latest · Tier 2 — Notable_

![v0.28.0
Highlights
This release features 584 commits from 270 contributors (76 new)!

Kimi-K3 performance push: a major optimization effort for Kimi-K3 across the stack — Decode Context Parallel (D... — Primary](https://opengraph.githubassets.com/4ebc6bdc52e9be13e062849c7d527d88f083515ccff086b45b804e1e708e9324/vllm-project/vllm/releases/tag/v0.28.0)

vLLM released version 0.28.0, adding disk-based KV-cache offloading, E/P/D disaggregation in Model Runner V2, and a Rust frontend with gRPC multimodal image inference. The project also raised the default maximum batched tokens from 8,192 to 16,384 and enabled prefix caching by default for Mamba models. It lists release wheels for CUDA, CPU and other targets, while moving bitsandbytes support to an out-of-tree plugin and removing several deprecated options.

## Sources

- [GitHub](https://github.com/vllm-project/vllm/releases/tag/v0.28.0)

---
Canonical: https://techandbusiness.org/newswire/Qm_v3PIK7C_g-kVhIq8Pub
Published: 2026-08-29T22:06:41.185Z
Story chronology: 2026-08-29T18:22:00.000Z
Retrieved: 2026-10-14T01:46:36.873Z
Publisher: Tech & Business (techandbusiness.org)