Pull Requests vllm-project/vllm

[Bugfix] Gemma 4: Fix bug around invalid JSON diffs during tool usage bug tool-calling

#38945 opened 2026-04-03 21:02 by Brummi

Re-enable Inductor pre-grad passes in standalone compile (torch>=2.12) ready

#38944 opened 2026-04-03 20:45 by frgossen

[BUG] Fix PP for R3 bug v1

#38943 opened 2026-04-03 20:42 by hao-aaron

[vLLM IR] Avoid redundant file reads in IrOpImpl.uuid() vllm-ir

#38940 opened 2026-04-03 19:45 by sBobHuang

[R3] Add routed experts to openai entrypoint frontend

#38939 opened 2026-04-03 19:25 by hao-aaron

Bug/test eagle dp v0 bug ready v1

#38938 opened 2026-04-03 19:19 by Monishver11

[ROCm][CI] Added back missing common deps rocm ready ci/build

#38937 opened 2026-04-03 19:16 by AndreasKaratzas

[PD][HeteroArch]Fix accuracy issue with CPU_ATTN as Decoder and Flash_ATTN as prefiller intel-gpu cpu kv-connector

#38935 opened 2026-04-03 18:58 by xuechendi

[Performance Improvement] Update `batched_count_greater_than` to handle batch size 1 without recompile v1

#38933 opened 2026-04-03 18:56 by Lucaskabela

[Bugfix][Perf] Indexer upcast WK to BF16 for fusion bug deepseek

#38928 opened 2026-04-03 18:06 by benchislett

[Bugfix][LoRA] Fix missing in_proj_z in Qwen3_5ForConditionalGenerati… bug ready qwen

#38927 opened 2026-04-03 17:59 by elenalil-aws

[Bugfix] Fix broken explicit unquantized kv cache dtype support bug

#38922 opened 2026-04-03 16:22 by Isotr0py

[ROCm][CI] Move skipped tests out of run-amd-test.sh rocm ci/build

#38921 opened 2026-04-03 16:16 by micah-wil

[Docs] add cache directory security guidance documentation

#38920 opened 2026-04-03 15:44 by russellb

[Bugfix] Runtime driver check for cuMemcpyBatchAsync in swap_blocks_batch bug

#38919 opened 2026-04-03 15:18 by Etelis

[Bugfix] Fix Qwen3.5 LoRA activation for shared expert modules bug qwen

#38917 opened 2026-04-03 14:52 by jayden222

[Bug] Fix compile error for `swap_blocks_batch` in CUDA 13 bug ready nvidia

#38915 opened 2026-04-03 13:58 by yewentao256

[ROCm] mi250x decode regression rocm

#38914 opened 2026-04-03 13:37 by rlrs

[NVIDIA] Update FlashInfer to version 0.6.7.post1. Avoid re-downloading BMM export headers when flashinfer-cubin is installed ready ci/build nvidia ready-run-all-tests

#38913 opened 2026-04-03 11:34 by johnnynunez

[Bugfix][Frontend] Fix Gemma4 streaming HTML duplication after tool calls bug tool-calling

#38909 opened 2026-04-03 11:03 by yoke233

Fix the order of _free_encoder_inputs v1

#38907 opened 2026-04-03 10:40 by gty111

[Core] Make handshake timeout configurable v1

#38906 opened 2026-04-03 10:10 by foraxe

refactor hard coded device string in test files under tests/compile tests/quantization tests/models and tests/model_executor v1 multi-modality

#38901 opened 2026-04-03 09:30 by wincent8

[P/D][Feature] Support kv_transfer_params for parallel sampling (n>1) frontend v1 kv-connector

#38900 opened 2026-04-03 09:24 by chaunceyjiang

[XPU] [CT] Enable CT W4A4MxFp4 path and add xpu kernel intel-gpu

#38896 opened 2026-04-03 08:36 by zufangzhu

bugfix(flashinfer,dcp): remove kv_cache_layout for BatchDCPPrefillWrapper._new_tokens. bug v1 nvidia

#38895 opened 2026-04-03 08:35 by pisceskkk

[Gemma4] Allow per-layer attention backend selection for heterogeneou…

#38891 opened 2026-04-03 07:48 by CunXin1

[Bugfix] Fix logger.warning format string arg mismatch in Qwen3XML tool parser bug tool-calling qwen

#38890 opened 2026-04-03 07:44 by edenfunf

[Doc] add Apple Clang 21+ compilation troubleshooting documentation cpu

#38889 opened 2026-04-03 07:40 by Alex-ai-future

[Multimodal] Add attention-score-based image token pruning for Qwen VL models documentation frontend v1 multi-modality qwen

#38888 opened 2026-04-03 07:00 by shhn1