Repository projects

• NVIDIA

  PRs and issues related to NVIDIA hardware

  #31 updated Jul 30, 2026

• Structured Output

  #16 updated Jul 30, 2026

• Transformers modeling backend

  Work on the Transformers modeling backend: running Transformers model implementations inside vLLM.

  #28 updated Jul 30, 2026

• CI Failures

  Tracking failures that are occurring in CI.

  #20 updated Jul 29, 2026

• CPU

  Track CPU related issues & tasks

  #42 updated Jul 29, 2026

• Prometheus Metrics

  Maintainer's tracking board for Prometheus metrics related PRs and issues

  #44 updated Jul 29, 2026

• torch.compile integration

  torch.compile integration related

  #12 updated Jul 29, 2026

• Ray

  Tracks Ray issues and pull requests in vLLM

  #7 updated Jul 28, 2026

• DeepSeek V3/R1

  2025-02-25: DeepSeek V3/R1 is supported with optimized block FP8 kernels, MLA, MTP spec decode, multi-node PP, EP, and W4A16 quantization

  #5 updated Jul 28, 2026

• Multi-modality Core

  Main tasks for the multi-modality workstream (#4194)

  #8 updated Jul 27, 2026

• Qwen3.5

  Optimization and bugfixes for Qwen3.5 model series.

  #50 updated Jul 26, 2026

• Large-Scale Serving

  #47 updated Jul 26, 2026

• Multi-modal Model Requests

  Community requests for multi-modal models

  #10 updated Jul 25, 2026

• Ongoing Sprints

  #57 updated Jul 25, 2026

• Startup Time

  #24 updated Jul 24, 2026

• vLLM IR

  #51 updated Jul 24, 2026

• DSR1 NVFP4 Model Bash

  #46 updated Jul 16, 2026

• Batch-invariant Inference

  #29 updated Jul 14, 2026

• Docker Image Optimization

  #33 updated Jun 22, 2026

• Onboarding Tasks

  A list of onboarding tasks for first-time contributors to get started with vLLM.

  #6 updated May 31, 2026

• MoE Refactor

  #45 updated May 17, 2026

• CI

  Backlog for CI feature requests

  #35 updated May 8, 2026

• V0 Deprecation

  #25 updated Mar 7, 2026

• Llama Issues & Bugs

  Tracker of known issues and bugs for serving Llama on vLLM

  #14 updated Feb 6, 2026

• Llama Features & Optimizations

  Enhancement to Llama herd of models. See also https://github.com/vllm-project/vllm/issues/16114

  #13 updated Nov 20, 2025

• [V1] Pipeline Parallelism

  [Testing] Optimize V1 PP efficiency.

  #1 updated Oct 6, 2025

• [V1] Speculative Decoding

  #2 updated Aug 15, 2025