Pull Requests vllm-project/vllm

[Misc] cleanup not use customer ops copy_blocks and copy_blocks_mla

#30967 opened 2025-12-18 14:02 by lengrongfu

Migrate some old models to Transformers modelling backend documentation new-model

#30966 opened 2025-12-18 13:51 by hmellor

[Perf][ROCm][AWQ] Improve performance of fused MoE GPTQ-AWQ and AWQ dequant kernels rocm needs-rebase

#30965 opened 2025-12-18 13:05 by yuttian1

[Quantization] support logical_widths for fp8 marlin

#30962 opened 2025-12-18 12:36 by jinzhen-lin

[Quantization] fix marlin w8a8 check

#30961 opened 2025-12-18 12:21 by jinzhen-lin

Migrate from `mypy` to `ty` ci/build

#30960 opened 2025-12-18 12:04 by hmellor

[Feature]: Support NVIDIA ModelOpt HF FP8 variants FP8_PER_CHANNEL_PER_TOKEN and FP8_PB_WO in vLLM documentation frontend nvidia

#30957 opened 2025-12-18 09:46 by CedricHwong

Hotfix container build documentation ci/build cpu

#30953 opened 2025-12-18 09:03 by maryamtahhan

[Misc] Improve worker error messages for better debugging v1 cpu

#30951 opened 2025-12-18 08:42 by yurekami

[Metrics] Add Prometheus counters for Model FLOPs Utilization (MFU) documentation ready v1

#30950 opened 2025-12-18 08:41 by markmc

[Doc] Add Sophgo TPU Support documentation

#30949 opened 2025-12-18 08:30 by wzyrrr

fix: Suppress torch.frombuffer UserWarning for non-writable buffers

#30948 opened 2025-12-18 08:25 by yurekami

[RFC][docs] Add lightweight AI assisted contribution policy documentation

#30947 opened 2025-12-18 08:22 by markmc

[Core] Improve DP synchronization error messages v1

#30946 opened 2025-12-18 08:10 by yurekami

[XPU] enable fp8 online streaming quantization

#30944 opened 2025-12-18 07:44 by yma11

[Feature] Add --ssl-ciphers CLI argument for TLS cipher control frontend

#30937 opened 2025-12-18 07:03 by rickychen-infinirc

[v1][CP] Improve DCP/PCP/MTP error messages with actionable guidance v1

#30936 opened 2025-12-18 06:59 by yurekami

[XPU] allow custom workers (e.g. vllm-omni workers) to be used on XPU

#30935 opened 2025-12-18 06:56 by faaany

enable vllm ut for Intel GPU documentation rocm speculative-decoding v1 nvidia

#30932 opened 2025-12-18 06:06 by wincent8

[Multimodal] Add FIPS 140-3 compliant hashing support multi-modality

#30925 opened 2025-12-18 05:07 by yurekami

[BugFix] Fix TypeError: unhashable type: 'dict' when serving deepseek32 ready v1 deepseek

#30924 opened 2025-12-18 04:16 by LucasWilkinson

[Feature] Support using FlexKV as anothor KV Cache Offloading option. documentation needs-rebase v1 kv-connector

#30917 opened 2025-12-18 02:35 by axxx03

[BugFix] Fix spec decode + structured outputs + preemption edge case bug ready v1

#30916 opened 2025-12-18 01:58 by njhill

[Bug] Fix torch inductor issue ready qwen

#30914 opened 2025-12-18 00:35 by yewentao256

[docker] install cuda13 version of lmcache and nixl ci/build kv-connector nvidia

#30913 opened 2025-12-17 23:47 by soodoshll

Fix/get raw stream patch #30905

#30912 opened 2025-12-17 23:39 by baonudesifeizhai

Migrate activation kernels to libtorch stable ABI ci/build cpu nvidia

#30908 opened 2025-12-17 22:49 by mikaylagawarecki

[Bug] Fix batch invariant in torch 2.10 ready

#30907 opened 2025-12-17 22:20 by yewentao256

[Feature]: support serving nvfp4 W4A16 moe models uisng Marlin

#30906 opened 2025-12-17 21:59 by EdalatiAli

custom build backend rocm needs-rebase ci/build

#30901 opened 2025-12-17 19:47 by dtrifiro