# vLLM v0.6.4 — Release & Upgrade Intelligence
- **Product:** vLLM (https://vllm.ai)
- **Version:** v0.6.4
- **Released:** Nov 15, 2024
- **Impact Score:** 5/10
- **Breaking Changes:** Not verified
- **Canonical Analysis:** https://changelogs.us/vllm/updates/v0.6.4/

## Executive Summary (TL;DR)
vLLM v0.6.4. Source highlight: Significant progress in V1 engine core refactor (#9826, #10135, #10288, #10211, #10225, #10228, #10268, #9954, #10272, #9971, #10224, #10166, #9289, #10058, #9888, #9972, #10059, #9945, #9679, #9871, #10227, #10245, #9629, #10097, #10203, #10148). You can checkout more details regarding the design a

## New Features
- Add user-configurable `--task` parameter for models that support both generation and embedding (#9424)
- Gaudi: Add Intel Gaudi (HPU) inference backend (#6143)
- CPU: Add embedding models support for CPU backend (#10193)
- Triton: Add Triton implementation for scaled_mm_triton to support fp8 and int8 SmoothQuant, symmetric case (#9857)
- Add goodput metric support (#9338)
- Add stateless process group for easier integration with RLHF and disaggregated prefill (#10216, #10072)

## Improvements
- Combine chunked prefill with speculative decoding (#9291)
- `fused_moe` Performance Improvement (#9384)
- Improvements to the pull request experience with DCO, mergify, stale bot, etc. (#9436, #9512, #9513, #9259, #10082, #10285, #9803)
- [Kernel][Model] Improve continuous batching for Jamba and Mamba
- [Performance][Spec Decode] Optimize ngram lookup performance
- [Performance][Kernel] Fused_moe Performance Improvement

## Bug Fixes
- [TPU] Fix TPU SMEM OOM by Pallas paged attention kernel
- [Bugfix] Clean up some cruft in mamba.py
- [Doc] Fix code formatting in spec_decode.rst
- [Bugfix] Update InternVL input mapper to support image embeds
- [BugFix] Fix chat API continuous usage stats
- [Bugfix] Fix vLLM UsageInfo and logprobs None AssertionError with empty token_ids

## Upgrade Command
```bash
python -m pip install 'vllm==0.6.4'
```