Pull Requests vllm-project/vllm

[torch.compile] Use FakeTensors instead of real GPU tensors for single-size compilation

#36093 opened 2026-03-05 04:58 by zou3519

[ROCm] Fix AITER ops fake impl and minor bugs rocm

#36092 opened 2026-03-05 04:57 by ChuanLi1101

[ROCm][CI] Making some tests optional to reduce workload rocm ci/build

#36090 opened 2026-03-05 04:46 by AndreasKaratzas

[Bugfix] Handle TimeoutError in Voxtral buffer_realtime_audio to prevent silent hang bug

#36089 opened 2026-03-05 04:32 by OiPunk

Don't fire ray compatibility webhook when PR or branch is not provided ci/build

#36088 opened 2026-03-05 04:22 by jeffreywang-anyscale

[AMD][Build] Add DeepEP to ROCm Dockerfile rocm ci/build

#36086 opened 2026-03-05 04:15 by rjrock

[Hardware] Replace `torch.cuda.synchronize()` api with `torch.accelerator.synchronize` documentation performance ready v1 nvidia ready-run-all-tests

#36085 opened 2026-03-05 03:57 by jikunshang

Add adaptive decode chunking for SM100 fused TRTLLM path (TMP FIX)#34988 v1 nvidia

#36083 opened 2026-03-05 03:46 by baonudesifeizhai

Perf: Optimize DeepEP prepare/finalize for identity mapping

#36081 opened 2026-03-05 03:11 by xueliangyang-oeuler

[PluggableLayer][4/N] Apply PluggableLayer to remaining layers

#36080 opened 2026-03-05 03:09 by whx-sjtu

Yejin/bench sleep wake timeout performance frontend tpu needs-rebase v1 cpu

#36079 opened 2026-03-05 02:59 by YJYJLee

Enable ModelRunnerV2 on XPU v1

#36078 opened 2026-03-05 02:11 by xinyu-intel

Revert "[Hardware] Replace `torch.cuda.empty_cache` with `torch.accelerator.empty_cache`" (#30681) documentation performance structured-output v1 nvidia

#36076 opened 2026-03-05 02:05 by zhewenl

[Docs] Add doc note about building for free-threaded Python. documentation

#36074 opened 2026-03-05 01:45 by nascheme

[WIP][Proof of concept] Overlap model loading and torch.compile frontend v1

#36072 opened 2026-03-05 01:35 by zou3519

[Bugfix][DCP] Fix CUDA graph capture for Decode Context Parallelism bug v1 nvidia

#36070 opened 2026-03-05 01:01 by sungsooha

fix(lora): bounds-check lora_a/lora_b in MergedColumnParallelLinear.set_lora

#36069 opened 2026-03-05 00:43 by JackYoung27

[Bugfix] Allow inherited_fds to be None to fix warnings when using spawn bug v1

#36068 opened 2026-03-05 00:30 by tjohnson31415

set VLLM_USE_BYTECODE_HOOK to 0 by default

#36067 opened 2026-03-05 00:26 by laithsakka

test Qwen/Qwen3-4B-Instruct-2507 for unbacked qwen

#36064 opened 2026-03-05 00:00 by laithsakka

[Refactor] Consolidate SupportsEagle v1 llama qwen gpt-oss kv-connector

#36063 opened 2026-03-04 23:54 by benchislett

[Kernel] [Helion] [11/N] Retune configs for silu_mul_fp8 ready

#36062 opened 2026-03-04 23:51 by gmagogsfm

[Bugfix] Fix DP/EP Shared Expert With Monolithic Kernels bug

#36061 opened 2026-03-04 23:51 by robertgshaw2-redhat

fix: force prefill path for MTP drafting on SM121 (GB10 Spark) v1 nvidia

#36060 opened 2026-03-04 23:34 by scottgl9

[BugFix] Fallback from FA4->FA2 for Batch Invariance bug v1

#36059 opened 2026-03-04 23:08 by frankwang28

[2/n] Migrate per_token_group_quant to torch stable ABI ci/build nvidia

#36058 opened 2026-03-04 23:06 by mikaylagawarecki

Adding deterministic lora benchmarking to vLLM Bench performance

#36057 opened 2026-03-04 22:58 by RonaldBXu

[Bugfix] Fix Deepseekv32 tool parser when stream interval > 1 bug deepseek

#36056 opened 2026-03-04 22:48 by sfeng33

[Bugfix] Fix zombie EngineCore processes after parent exit bug v1

#36055 opened 2026-03-04 22:46 by AjAnubolu

[Bugfix] Fix tokenize endpoint malformed token_strs bug frontend

#36054 opened 2026-03-04 22:41 by AjAnubolu