# vLLM v0.4.2 — Release & Upgrade Intelligence
- **Product:** vLLM (https://vllm.ai)
- **Version:** v0.4.2
- **Released:** May 05, 2024
- **Impact Score:** 5/10
- **Breaking Changes:** Not verified
- **Canonical Analysis:** https://changelogs.us/vllm/updates/v0.4.2/

## Executive Summary (TL;DR)
vLLM v0.4.2. Source highlight: Chunked prefill is ready for testing! It improves inter-token latency in high load scenario by chunking the prompt processing and priortizes decode (#4580)

## New Features
- Chunked prefill is ready for testing! It improves inter-token latency in high load scenario by chunking the prompt processing and priortizes decode (#4580)
- Speculative decoding functionalities: logprobs (#4378), ngram (#4237)
- Support FlashInfer as attention backend (#4353)
- Add support for Phi-3-mini (#4298, #4372, #4380)
- Add more histogram metrics (#2764, #4523)
- [Doc] Add note for docker user

## Improvements
- Full tensor parallelism for LoRA layers (#3524)
- Expanding Marlin kernel to support all GPTQ models (#3922, #4466, #4533)

## Bug Fixes
- [Bugfix] Fix marlin kernel crash on H100
- [Bugfix][Model] Refactor OLMo model to support new HF format in transformers 4.40.0
- [Doc] README Phi-3 name fix.
- [Bugfix] Fix parameter name in `get_tokenizer`
- [Frontend][Bugfix] Disallow extra fields in OpenAI API
- [Misc] Fix logger format typo

## Upgrade Command
```bash
python -m pip install 'vllm==0.4.2'
```