# vLLM v0.9.2 — Release & Upgrade Intelligence
- **Product:** vLLM (https://vllm.ai)
- **Version:** v0.9.2
- **Released:** Jul 07, 2025
- **Impact Score:** 5/10
- **Breaking Changes:** Not verified
- **Canonical Analysis:** https://changelogs.us/vllm/updates/v0.9.2/

## Executive Summary (TL;DR)
vLLM v0.9.2. Source highlight: This release contains 452 commits from 167 contributors (31 new!)

## New Features
- Expert‑Parallel Load Balancer (EPLB) has been added! (#18343, #19790, #19885).
- SM120: CUTLASS W8A8/FP8 kernels and related tuning, added to Dockerfile (#17280, #19566, #20071, #19794)
- Add models and features supporting matrix. (#20230)
- Add H20-3e fused MoE kernel tuning configs for Qwen3-235B-A22B-FP8
- Add fused MOE config for Qwen3 30B A3B on B200
- [ROCm] Add rules to automatically label ROCm related PRs

## Improvements
- Shared `CachedRequestData` objects and cached sampler‑ID stores deliver perf enhancements (#20232, #20291).
- Disaggregated serving enhancements: Avoid stranding blocks in P when aborted in D's waiting queue (#19223), let toy proxy handle /chat/completions (#19730)
- Native xPyD P2P NCCL transport as a base case for native PD without external dependency (#18242, #20246).
- SM100: block‑scaled‑group GEMM, INT8/FP8 vectorization, deep‑GEMM kernels, activation‑chunking for MoE, and group‑size 64 for Machete (#19757, #19572, #19168, #19085, #20290, #20331).
- Intel GPU (V1) backend with Flash‑Attention support (#19560).
- AMD ROCm: full‑graph capture for TritonAttention, quick All‑Reduce, and chunked pre‑fill (#19158, #19744, #18596).

## Bug Fixes
- Full CUDA‑Graph execution is now available for all FlashAttention v3 (FA3) and FlashMLA paths, including prefix‑caching. CUDA graph now has a live capture progress bar makes debugging easier (#20301, #18581, #19617, #19501).
- TPU: dynamic‑grid KV‑cache updates, head‑dim less than 128, tuned paged‑attention kernels, and KV‑padding fixes (#19928, #20235, #19620, #19813, #20048, #20339).
- [BugFix] Allow use_cudagraph to work with dynamic VLLM_USE_V1
- [BugFix] Fix docker build cpu-dev image error
- Fix test_max_model_len in tests/entrypoints/llm/test_generate.py
- Fix Typo in Documentation and Function Name

## Upgrade Command
```bash
python -m pip install 'vllm==0.9.2'
```