vLLM 0.8.3 release notes

Highlights

This release features 260 commits, 109 contributors, 38 new contributors.

We are excited to announce Day 0 Support for Llama 4 Scout and Maverick (#16104). Please see our blog for detailed user guide.
- Please note that Llama4 is only supported in V1 engine only for now.
V1 engine now supports native sliding window attention (#14097) with the hybrid memory allocator.

Cluster Scale Serving

Single node data parallel with API server support (#13923)
Multi-node offline DP+EP example (#15484)
Expert parallelism enhancements
- CUTLASS grouped gemm fp8 MoE kernel (#13972)
- Fused experts refactor (#15914)
- Fp8 Channelwise Dynamic Per Token GroupedGEMM (#15587)
- Adding support for fp8 gemm layer input in fp8 (#14578)
- Add option to use DeepGemm contiguous grouped gemm kernel for fused MoE operations. (#13932)
Support XpYd disaggregated prefill with MooncakeStore (#12957)

Model Supports

Llama 4 (#16104), Aya Vision (#15441), MiniMaxText01(#13454), Skywork-R1V (#15397), jina-reranker-v2 (#15876)
Add Reasoning Parser for Granite Models (#14202)
Add Phi-4-mini function calling support (#14886)

V1 Engine

Collective RPC (#15444)
Faster top-k only implementation (#15478)
BitsAndBytes support (#15611)
Speculative Decoding: metrics (#15151), Eagle Proposer (#15729), n-gram interface update (#15750), EAGLE Architecture with Proper RMS Norms (#14990)

Features

API

Support Enum for xgrammar based structured output in V1. (#15594, #15757)
A new tags parameter for wake_up (#15500)
V1 LoRA support CPU offload (#15843)
Prefix caching support: FIPS enabled machines with MD5 hashing (#15299), SHA256 as alternative hashing algorithm (#15297)
Addition of http service metrics (#15657)

Performance

LoRA Scheduler optimization bridging V1 and V0 performance (#15422).

Hardwares

AMD:
- Add custom allreduce support for ROCM (#14125)
- Quark quantization documentation (#15861)
- AITER integration: int8 scaled gemm kernel (#15433), fused moe (#14967)
- Paged attention for V1 (#15720)
CPU:
- CPU MLA (#14744)
TPU
- Improve Memory Usage Estimation (#15671)
- Optimize the all-reduce performance (#15903)
- Support sliding window and logit soft capping in the paged attention kernel. (#15732)
- TPU-optimized top-p implementation (avoids scattering). (#15736)

Doc, Build, Ecosystem

V1 user guide update: fp8 kv cache support (#15585), multi-modality (#15460)
Recommend developing with Python 3.12 in developer guide (#15811)
Clean up: move dockerfiles into their own directory (#14549)
Add minimum version for huggingface_hub to enable Xet downloads (#15873)
TPU CI: Add basic perf regression test (#15414)

What's Changed

Fix CUDA kernel index data type in vllm/csrc/quantization/gptq_marlin/awq_marlin_repack.cu +10 by @houseroad in https://github.com/vllm-project/vllm/pull/15160
[Hardware][TPU][Bugfix] Fix v1 mp profiler by @lsy323 in https://github.com/vllm-project/vllm/pull/15409
[Kernel][CPU] CPU MLA by @gau-nernst in https://github.com/vllm-project/vllm/pull/14744
Dockerfile.ppc64le changes to move to UBI by @Shafi-Hussain in https://github.com/vllm-project/vllm/pull/15402
[Misc] Clean up MiniCPM-V/O code by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15337
[Misc] Remove redundant num_embeds by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15443
[Doc] Update V1 user guide for multi-modality by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15460
[Kernel] Fix conflicting macro names for gguf kernels by @SzymonOzog in https://github.com/vllm-project/vllm/pull/15456
[bugfix] fix inductor cache on max_position_embeddings by @youkaichao in https://github.com/vllm-project/vllm/pull/15436
[CI/Build] Add tests for the V1 tpu_model_runner. by @yarongmu-google in https://github.com/vllm-project/vllm/pull/14843
[Bugfix] Support triton==3.3.0+git95326d9f for RTX 5090 (Unsloth + vLLM compatibility) by @oteroantoniogom in https://github.com/vllm-project/vllm/pull/15471
[bugfix] add supports_v1 platform interface by @joerunde in https://github.com/vllm-project/vllm/pull/15417
Add workaround for shared field_names in pydantic model class by @maxdebayser in https://github.com/vllm-project/vllm/pull/13925
[TPU][V1] Fix Sampler recompilation by @NickLucche in https://github.com/vllm-project/vllm/pull/15309
[V1][Minor] Use SchedulerInterface type for engine scheduler field by @njhill in https://github.com/vllm-project/vllm/pull/15499
[V1] Support long_prefill_token_threshold in v1 scheduler by @houseroad in https://github.com/vllm-project/vllm/pull/15419
[core] add bucket padding to tpu_model_runner by @Chenyaaang in https://github.com/vllm-project/vllm/pull/14995
[Core] LoRA: V1 Scheduler optimization by @varun-sundar-rabindranath in https://github.com/vllm-project/vllm/pull/15422
[CI/Build] LoRA: Delete long context tests by @varun-sundar-rabindranath in https://github.com/vllm-project/vllm/pull/15503
Transformers backend already supports V1 by @hmellor in https://github.com/vllm-project/vllm/pull/15463
[Model] Support multi-image for Molmo by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15438
[Misc] Warn about v0 in benchmark_paged_attn.py by @tlrmchlsmth in https://github.com/vllm-project/vllm/pull/15495
[BugFix] Fix nightly MLA failure (FA2 + MLA chunked prefill, i.e. V1, producing bad results) by @LucasWilkinson in https://github.com/vllm-project/vllm/pull/15492
[misc] LoRA - Skip LoRA kernels when not required by @varun-sundar-rabindranath in https://github.com/vllm-project/vllm/pull/15152
Fix raw_request extraction in load_aware_call decorator by @daniel-salib in https://github.com/vllm-project/vllm/pull/15382
[Feature] Enhance EAGLE Architecture with Proper RMS Norms by @luyuzhe111 in https://github.com/vllm-project/vllm/pull/14990
[FEAT][ROCm] Integrate Fused MoE Kernels from AITER by @vllmellm in https://github.com/vllm-project/vllm/pull/14967
[Misc] Enhance warning information to user-defined chat template by @wwl2755 in https://github.com/vllm-project/vllm/pull/15408
[Misc] improve example script output by @reidliu41 in https://github.com/vllm-project/vllm/pull/15528
Separate base model from TransformersModel by @hmellor in https://github.com/vllm-project/vllm/pull/15467
Apply torchfix by @cyyever in https://github.com/vllm-project/vllm/pull/15532
Improve validation of TP in Transformers backend by @hmellor in https://github.com/vllm-project/vllm/pull/15540
[Model] Add Reasoning Parser for Granite Models by @alex-jw-brooks in https://github.com/vllm-project/vllm/pull/14202
multi-node offline DP+EP example by @youkaichao in https://github.com/vllm-project/vllm/pull/15484
Fix weight loading for some models in Transformers backend by @hmellor in https://github.com/vllm-project/vllm/pull/15544
[Refactor] Remove passthrough backend when generate grammar by @aarnphm in https://github.com/vllm-project/vllm/pull/15317
[V1][Sampler] Faster top-k only implementation by @njhill in https://github.com/vllm-project/vllm/pull/15478
Support SHA256 as hash function in prefix caching by @dr75 in https://github.com/vllm-project/vllm/pull/15297
Applying some fixes for K8s agents in CI by @Alexei-V-Ivanov-AMD in https://github.com/vllm-project/vllm/pull/15493
[V1] TPU - Revert to exponential padding by default by @alexm-redhat in https://github.com/vllm-project/vllm/pull/15565
[V1] TPU CI - Fix test_compilation.py by @alexm-redhat in https://github.com/vllm-project/vllm/pull/15570
Use Cache Hinting for fused_moe kernel by @wrmedford in https://github.com/vllm-project/vllm/pull/15511
[TPU] support disabling xla compilation cache by @yaochengji in https://github.com/vllm-project/vllm/pull/15567
Support FIPS enabled machines with MD5 hashing by @MattTheCuber in https://github.com/vllm-project/vllm/pull/15299
[Kernel] CUTLASS grouped gemm fp8 MoE kernel by @ElizaWszola in https://github.com/vllm-project/vllm/pull/13972
Add automatic tpu label to mergify.yml by @mgoin in https://github.com/vllm-project/vllm/pull/15560
add platform check back by @Chenyaaang in https://github.com/vllm-project/vllm/pull/15578
[misc] LoRA: Remove unused long context test data by @varun-sundar-rabindranath in https://github.com/vllm-project/vllm/pull/15558
[Doc] Update V1 user guide for fp8 kv cache support by @wayzeng in https://github.com/vllm-project/vllm/pull/15585
[moe][quant] add weight name case for offset by @MengqingCao in https://github.com/vllm-project/vllm/pull/15515
[V1] Refactor num_computed_tokens logic by @comaniac in https://github.com/vllm-project/vllm/pull/15307
Allow torchao quantization in SiglipMLP by @jerryzh168 in https://github.com/vllm-project/vllm/pull/15575
[ROCm] Env variable to trigger custom PA by @gshtras in https://github.com/vllm-project/vllm/pull/15557
[TPU] [V1] fix cases when max_num_reqs is set smaller than MIN_NUM_SEQS by @yaochengji in https://github.com/vllm-project/vllm/pull/15583
[Misc] Restrict ray version dependency and update PP feature warning in V1 by @ruisearch42 in https://github.com/vllm-project/vllm/pull/15556
[TPU] Avoid Triton Import by @robertgshaw2-redhat in https://github.com/vllm-project/vllm/pull/15589
[Misc] Consolidate LRUCache implementations by @Avabowler in https://github.com/vllm-project/vllm/pull/15481
[Quantization] Fp8 Channelwise Dynamic Per Token GroupedGEMM by @robertgshaw2-redhat in https://github.com/vllm-project/vllm/pull/15587
[Misc] Clean up scatter_patch_features by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15559
[Misc] Use model_redirect to redirect the model name to a local folder. by @noooop in https://github.com/vllm-project/vllm/pull/14116
Fix incorrect filenames in vllm_compile_cache.py by @zou3519 in https://github.com/vllm-project/vllm/pull/15494
[Doc] update --system for transformers installation in docker doc by @reidliu41 in https://github.com/vllm-project/vllm/pull/15616
[Model] MiniCPM-V/O supports V1 by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15487
[Bugfix] Fix use_cascade_attention handling for Alibi-based models on vllm/v1 by @h-sugi in https://github.com/vllm-project/vllm/pull/15211
[Doc] Link to onboarding tasks by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15629
[Misc] Replace is_encoder_decoder_inputs with split_enc_dec_inputs by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15620
[Feature] Add middleware to log API Server responses by @terrytangyuan in https://github.com/vllm-project/vllm/pull/15593
[Misc] Avoid direct access of global mm_registry in compute_encoder_budget by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15621
[Doc] Use absolute placement for Ask AI button by @hmellor in https://github.com/vllm-project/vllm/pull/15628
[Bugfix][TPU][V1] Fix recompilation by @NickLucche in https://github.com/vllm-project/vllm/pull/15553
Correct PowerPC to modern IBM Power by @clnperez in https://github.com/vllm-project/vllm/pull/15635
[CI] Update rules for applying tpu label. by @russellb in https://github.com/vllm-project/vllm/pull/15634
[V1] AsyncLLM data parallel by @njhill in https://github.com/vllm-project/vllm/pull/13923
[TPU] Lazy Import by @robertgshaw2-redhat in https://github.com/vllm-project/vllm/pull/15656
[Quantization][V1] BitsAndBytes support V1 by @jeejeelee in https://github.com/vllm-project/vllm/pull/15611
[Bugfix] Fix failure to launch in Tensor Parallel TP mode on macOS. by @kebe7jun in https://github.com/vllm-project/vllm/pull/14948
[Doc] Fix dead links in Job Board by @wwl2755 in https://github.com/vllm-project/vllm/pull/15637
[CI][TPU] Temporarily Disable Quant Test on TPU by @robertgshaw2-redhat in https://github.com/vllm-project/vllm/pull/15649
Revert "Use Cache Hinting for fused_moe kernel (#15511)" by @wrmedford in https://github.com/vllm-project/vllm/pull/15645
[Misc]add coding benchmark for speculative decoding by @CXIAAAAA in https://github.com/vllm-project/vllm/pull/15303
[Quantization][FP8] Adding support for fp8 gemm layer input in fp8 by @gshtras in https://github.com/vllm-project/vllm/pull/14578
Refactor error handling for multiple exceptions in preprocessing by @JasonZhu1313 in https://github.com/vllm-project/vllm/pull/15650
[Bugfix] Fix mm_hashes forgetting to be passed by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15668
[V1] Remove legacy input registry by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15673
[TPU][CI] Fix TPUModelRunner Test by @robertgshaw2-redhat in https://github.com/vllm-project/vllm/pull/15667
[Refactor][Frontend] Keep all logic about reasoning into one class by @gaocegege in https://github.com/vllm-project/vllm/pull/14428
[CPU][CI] Improve CPU Dockerfile by @bigPYJ1151 in https://github.com/vllm-project/vllm/pull/15690
[Bugfix] Fix 'InductorAdaptor object has no attribute 'cache_dir' by @jeejeelee in https://github.com/vllm-project/vllm/pull/15674
[Misc] Fix test_sleep to use query parameters by @lizzzcai in https://github.com/vllm-project/vllm/pull/14373
[Bugfix][Frontend] Eliminate regex based check in reasoning full generator by @gaocegege in https://github.com/vllm-project/vllm/pull/14821
[Frontend] update priority for --api-key and VLLM_API_KEY by @reidliu41 in https://github.com/vllm-project/vllm/pull/15588
[Docs] Add "Generation quality changed" section to troubleshooting by @hmellor in https://github.com/vllm-project/vllm/pull/15701
[Model] Adding torch compile annotations to chatglm by @jeejeelee in https://github.com/vllm-project/vllm/pull/15624
[Bugfix][v1] xgrammar structured output supports Enum. by @chaunceyjiang in https://github.com/vllm-project/vllm/pull/15594
[Bugfix] embed_is_patch for Idefics3 by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15696
[V1] Support disable_any_whtespace for guidance backend by @russellb in https://github.com/vllm-project/vllm/pull/15584
[doc] add missing imports by @reidliu41 in https://github.com/vllm-project/vllm/pull/15699
[Bugfix] Fix regex compile display format by @kebe7jun in https://github.com/vllm-project/vllm/pull/15368
Fix cpu offload testing for gptq/awq/ct by @mgoin in https://github.com/vllm-project/vllm/pull/15648
[Minor] Remove TGI launching script by @WoosukKwon in https://github.com/vllm-project/vllm/pull/15646
[Misc] Remove unused utils and clean up imports by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15708
[Misc] Remove stale func in KVTransferConfig by @ShangmingCai in https://github.com/vllm-project/vllm/pull/14746
[TPU] [Perf] Improve Memory Usage Estimation by @robertgshaw2-redhat in https://github.com/vllm-project/vllm/pull/15671
[Bugfix] [torch.compile] Add Dynamo metrics context during compilation by @ProExpertProg in https://github.com/vllm-project/vllm/pull/15639
[V1] TPU - Fix the chunked prompt bug by @alexm-redhat in https://github.com/vllm-project/vllm/pull/15713
[Misc] cli auto show default value by @reidliu41 in https://github.com/vllm-project/vllm/pull/15582
implement prometheus fast-api-instrumentor for http service metrics by @daniel-salib in https://github.com/vllm-project/vllm/pull/15657
[Docs][V1] Optimize diagrams in prefix caching design by @simpx in https://github.com/vllm-project/vllm/pull/15716
[ROCm][AMD][Build] Update AMD supported arch list by @gshtras in https://github.com/vllm-project/vllm/pull/15632
[Model] Support Skywork-R1V by @pengyuange in https://github.com/vllm-project/vllm/pull/15397
[Docs] Document v0 engine support in reasoning outputs by @gaocegege in https://github.com/vllm-project/vllm/pull/15739
[Misc][V1] Misc code streamlining by @njhill in https://github.com/vllm-project/vllm/pull/15723
[Bugfix] LoRA V1: add and fix entrypoints tests by @varun-sundar-rabindranath in https://github.com/vllm-project/vllm/pull/15715
[CI] Speed up V1 structured output tests by @russellb in https://github.com/vllm-project/vllm/pull/15718
Use numba 0.61 for python 3.10+ to support numpy>=2 by @cyyever in https://github.com/vllm-project/vllm/pull/15692
[Bugfix] set VLLM_WORKER_MULTIPROC_METHOD=spawn for vllm.entrypoionts.openai.api_server by @jinzhen-lin in https://github.com/vllm-project/vllm/pull/15700
[TPU][V1][Bugfix] Fix w8a8 recompiilation with GSM8K by @NickLucche in https://github.com/vllm-project/vllm/pull/15714
[Kernel][TPU][ragged-paged-attn] vLLM code change for PR#8896 by @yarongmu-google in https://github.com/vllm-project/vllm/pull/15659
[doc] update doc by @reidliu41 in https://github.com/vllm-project/vllm/pull/15740
[FEAT] [ROCm] Add AITER int8 scaled gemm kernel by @tjtanaa in https://github.com/vllm-project/vllm/pull/15433
[V1] [Feature] Collective RPC by @wwl2755 in https://github.com/vllm-project/vllm/pull/15444
[Feature][Disaggregated] Support XpYd disaggregated prefill with MooncakeStore by @ShangmingCai in https://github.com/vllm-project/vllm/pull/12957
[V1] Support interleaved modality items by @ywang96 in https://github.com/vllm-project/vllm/pull/15605
[V1][Minor] Simplify rejection sampler's parse_output by @WoosukKwon in https://github.com/vllm-project/vllm/pull/15741
[Bugfix] Fix Mllama interleaved images input support by @Isotr0py in https://github.com/vllm-project/vllm/pull/15564
[CI] xgrammar structured output supports Enum. by @chaunceyjiang in https://github.com/vllm-project/vllm/pull/15757
[Bugfix] Fix Mistral guided generation using xgrammar by @juliendenize in https://github.com/vllm-project/vllm/pull/15704
[doc] update conda to usage link in installation by @reidliu41 in https://github.com/vllm-project/vllm/pull/15761
fix test_phi3v by @pansicheng in https://github.com/vllm-project/vllm/pull/15321
[V1] Override mm_counts for dummy data creation by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15703
fix: lint fix a ruff checkout syntax error by @yihong0618 in https://github.com/vllm-project/vllm/pull/15767
[Bugfix] Added embed_is_patch mask for fuyu model by @kylehh in https://github.com/vllm-project/vllm/pull/15731
fix: Comments to English for better dev experience by @yihong0618 in https://github.com/vllm-project/vllm/pull/15768
[V1][Scheduler] Avoid calling _try_schedule_encoder_inputs for every request by @WoosukKwon in https://github.com/vllm-project/vllm/pull/15778
[Misc] update the comments by @lcy4869 in https://github.com/vllm-project/vllm/pull/15780
[Benchmark] Update Vision Arena Dataset and HuggingFaceDataset Setup by @JenZhao in https://github.com/vllm-project/vllm/pull/15748
[Feature][ROCm]Enable fusion pass for torch.compile on ROCm by @charlifu in https://github.com/vllm-project/vllm/pull/15050
Recommend developing with Python 3.12 in developer guide by @hmellor in https://github.com/vllm-project/vllm/pull/15811
fix: better install requirement for install in setup.py by @yihong0618 in https://github.com/vllm-project/vllm/pull/15796
[V1] Fully Transparent Implementation of CPU Offloading by @youkaichao in https://github.com/vllm-project/vllm/pull/15354
[Model] Update support for NemotronNAS models by @Naveassaf in https://github.com/vllm-project/vllm/pull/15008
[Bugfix] Fix Crashing When Loading Modules With Batchnorm Stats by @alex-jw-brooks in https://github.com/vllm-project/vllm/pull/15813
[Bugfix] Fix missing return value in load_weights method of adapters.py by @noc-turne in https://github.com/vllm-project/vllm/pull/15542
Upgrade transformers to v4.50.3 by @hmellor in https://github.com/vllm-project/vllm/pull/13905
[Bugfix] Check dimensions of multimodal embeddings in V1 by @DarkLight1337 in https://github.com/vllm-project/vllm/pull/15816
[V1][Spec Decode] Remove deprecated spec decode config params by @ShangmingCai in https://github.com/vllm-project/vllm/pull/15466
fix: change GB to GiB in logging close #14979 by @yihong0618 in https://github.com/vllm-project/vllm/pull/15807
[V1] TPU CI - Add basic perf regression test by @alexm-redhat in https://github.com/vllm-project/vllm/pull/15414
Fix Transformers backend compatibility check by @hmellor in https://github.com/vllm-project/vllm/pull/15290
[V1][Core] Remove unused speculative config from scheduler by @markmc in https://github.com/vllm-project/vllm/pull/15818
Move dockerfiles into their own directory by @hmellor in https://github.com/vllm-project/vllm/pull/14549
[Distributed] Add custom allreduce support for ROCM by @ilmarkov in https://github.com/vllm-project/vllm/pull/14125
Rename fallback model and refactor supported models section by @hmellor in https://github.com/vllm-project/vllm/pull/15829
[Frontend] Add Phi-4-mini function calling support by @kinfey in https://github.com/vllm-project/vllm/pull/14886
[Bugfix][Model] fix mllama multi-image by @yma11 in https://github.com/vllm-project/vllm/pull/14883
[Bugfix] Fix extra comma by @haochengxia in https://github.com/vllm-project/vllm/pull/15851
[Bugfix]: Fix is_embedding_layer condition in VocabParallelEmbedding by @alexwl in https://github.com/vllm-project/vllm/pull/15824
[V1] TPU - Fix fused MOE by @alexm-redhat in https://github.com/vllm-project/vllm/pull/15834
[sleep mode] clear pytorch cache after sleep by @lionelvillard in https://github.com/vllm-project/vllm/pull/15248
[ROCm] Use device name in the warning by @gshtras in https://github.com/vllm-project/vllm/pull/15838
[V1] Implement sliding window attention in kv_cache_manager by @heheda12345 in https://github.com/vllm-project/vllm/pull/14097
fix: can not use uv run collect_env close #13888 by @yihong0618 in https://github.com/vllm-project/vllm/pull/15792
[Feature] specify model in config.yaml by @wayzeng in https://github.com/vllm-project/vllm/pull/15798
[Misc] Enable V1 LoRA by default by @varun-sundar-rabindranath in https://github.com/vllm-project/vllm/pull/15320
[Misc] Fix speculative config repr string by @ShangmingCai in https://github.com/vllm-project/vllm/pull/15860
[Docs] Fix small error in link text by @hmellor in https://github.com/vllm-project/vllm/pull/15868
[Bugfix] Fix no video/image profiling edge case for MultiModalDataParser by @Isotr0py in https://github.com/vllm-project/vllm/pull/15828
[Misc] Use envs.VLLM_USE_RAY_COMPILED_DAG_CHANNEL_TYPE by @ruisearch42 in https://github.com/vllm-project/vllm/pull/15831
setup correct nvcc version with CUDA_HOME by @chenyang78 in https://github.com/vllm-project/vllm/pull/15725
[Model] Support Mistral3 in the HF Transformers format by @mgoin in https://github.com/vllm-project/vllm/pull/15505
[Misc] remove unused script by @reidliu41 in https://github.com/vllm-project/vllm/pull/15746
Remove format.sh as it's been unsupported >70 days by @hmellor in https://github.com/vllm-project/vllm/pull/15884
[New Model]: jinaai/jina-reranker-v2-base-multilingual by @noooop in https://github.com/vllm-project/vllm/pull/15876
[Doc] Quark quantization documentation by @cha557 in https://github.com/vllm-project/vllm/pull/15861
Reinstate format.sh and make pre-commit installation simpler by @hmellor in https://github.com/vllm-project/vllm/pull/15890
[Misc] Allow using OpenCV as video IO fallback by @Isotr0py in https://github.com/vllm-project/vllm/pull/15055
[ROCm][Build][Bugfix] Bring the base dockerfile in sync with the ROCm fork by @gshtras in https://github.com/vllm-project/vllm/pull/15820
Add option to use DeepGemm contiguous grouped gemm kernel for fused MoE operations. by @bnellnm in https://github.com/vllm-project/vllm/pull/13932
[CI/Build] Clean up LoRA tests by @jeejeelee in https://github.com/vllm-project/vllm/pull/15867
[Model] Aya Vision by @JenZhao in https://github.com/vllm-project/vllm/pull/15441
[Model] Add module name prefixes to gemma3 by @cloud11665 in https://github.com/vllm-project/vllm/pull/15889
[CI] Disable flaky structure decoding test temporarily. by @ywang96 in https://github.com/vllm-project/vllm/pull/15892
[V1][Metrics] Initial speculative decoding metrics by @markmc in https://github.com/vllm-project/vllm/pull/15151
[V1][Spec Decode] Implement Eagle Proposer [1/N] by @WoosukKwon in https://github.com/vllm-project/vllm/pull/15729
[Docs] update usage stats language by @simon-mo in https://github.com/vllm-project/vllm/pull/15898
[BugFix] make sure socket close by @yihong0618 in https://github.com/vllm-project/vllm/pull/15875
[Model][MiniMaxText01] Support MiniMaxText01 model inference by @ZZBoom in https://github.com/vllm-project/vllm/pull/13454
[Docs] Add Ollama meetup slides by @simon-mo in https://github.com/vllm-project/vllm/pull/15905
[Docs] Add Intel as Sponsor by @simon-mo in https://github.com/vllm-project/vllm/pull/15913
Fix input triton kernel for eagle by @ekagra-ranjan in https://github.com/vllm-project/vllm/pull/15909
[V1] Fix: make sure k_index is int64 for apply_top_k_only by @b8zhong in https://github.com/vllm-project/vllm/pull/15907
[Bugfix] Fix imports for MoE on CPU by @gau-nernst in https://github.com/vllm-project/vllm/pull/15841
[V1][Minor] Enhance SpecDecoding Metrics Log in V1 by @WoosukKwon in https://github.com/vllm-project/vllm/pull/15902
[Doc] Update rocm.inc.md by @chun37 in https://github.com/vllm-project/vllm/pull/15917
[V1][Bugfix] Fix typo in MoE TPU checking by @ywang96 in https://github.com/vllm-project/vllm/pull/15927
[Benchmark]Fix error message by @Potabk in https://github.com/vllm-project/vllm/pull/15866
[Misc] Replace print

These notes run past the length kept in the archive. The rest is on the publisher’s page.