●Change8

Migrating to vLLM v0.8.3

Version v0.8.3 introduces 2 breaking changes. This guide details how to update your code.

Released: 4/6/2025

2
Breaking Changes
4
Migration Steps
9
Affected Symbols

⚠️ Check Your Code

If you use any of these symbols, you need to read this guide:

vllm.v1.engineMooncakeStoreSchedulerInterfaceTransformersModelpydantic.v1vllm.csrc.quantization.gptq_marlinAITERDeepGemmxgrammar

Breaking Changes

●Issue #1

Llama 4 support is currently restricted to the V1 engine only.

●Issue #2

The n-gram interface for speculative decoding has been updated, which may require changes to custom implementations.

Migration Steps

  1. 1
    Upgrade huggingface_hub to the minimum required version to enable Xet downloads.
  2. 2
    If using Llama 4, ensure the V1 engine is enabled.
  3. 3
    Update developer environments to Python 3.12 as per the new recommendation.
  4. 4
    Review n-gram interface implementations if using speculative decoding to align with the updated interface.

Release Summary

This release introduces Day 0 support for Llama 4 (V1 engine only) and native sliding window attention. It also features significant performance optimizations for MoE kernels, expanded hardware support for AMD and TPU, and architectural improvements to the V1 engine.

Need More Details?

View the full release notes and all changes for vLLM v0.8.3.

View Full Changelog