pyproject.toml

[build-system]
# Should be mirrored in requirements/build.txt
requires = [
    "cmake>=3.26.1",
    "ninja",
    "packaging>=24.2",
    "setuptools>=77.0.3,<80.0.0",
    "setuptools-scm>=8.0",
    "torch == 2.8.0",
    "wheel",
    "jinja2",
]
build-backend = "setuptools.build_meta"

[project]
name = "vllm"
authors = [{name = "vLLM Team"}]
license = "Apache-2.0"
license-files = ["LICENSE"]
readme = "README.md"
description = "A high-throughput and memory-efficient inference and serving engine for LLMs"
classifiers = [
    "Programming Language :: Python :: 3.9",
    "Programming Language :: Python :: 3.10",
    "Programming Language :: Python :: 3.11",
    "Programming Language :: Python :: 3.12",
    "Programming Language :: Python :: 3.13",
    "Intended Audience :: Developers",
    "Intended Audience :: Information Technology",
    "Intended Audience :: Science/Research",
    "Topic :: Scientific/Engineering :: Artificial Intelligence",
    "Topic :: Scientific/Engineering :: Information Analysis",
]
requires-python = ">=3.9,<3.14"
dynamic = [ "version", "dependencies", "optional-dependencies"]

[project.urls]
Homepage="https://github.com/vllm-project/vllm"
Documentation="https://docs.vllm.ai/en/latest/"
Slack="https://slack.vllm.ai/"

[project.scripts]
vllm = "vllm.entrypoints.cli.main:main"

[project.entry-points."vllm.general_plugins"]
lora_filesystem_resolver = "vllm.plugins.lora_resolvers.filesystem_resolver:register_filesystem_resolver"

[tool.setuptools_scm]
# no extra settings needed, presence enables setuptools-scm

[tool.setuptools.packages.find]
where = ["."]
include = ["vllm*"]

[tool.ruff.lint.per-file-ignores]
"vllm/third_party/**" = ["ALL"]
"vllm/version.py" = ["F401"]
"vllm/_version.py" = ["ALL"]
# TEMPORARY! These ignores will be fixed forward
## Line length violations
"csrc/cutlass_extensions/vllm_cutlass_library_extension.py" = ["E501"]
"tests/compile/piecewise/test_simple.py" = ["E501"]
"tests/compile/piecewise/test_toy_llama.py" = ["E501", "B023"]
"tests/entrypoints/conftest.py" = ["E501"]
"tests/entrypoints/openai/test_audio.py" = ["E501"]
"tests/entrypoints/openai/test_chat.py" = ["E501"]
"tests/entrypoints/openai/test_chat_template.py" = ["E501"]
"tests/entrypoints/openai/test_chat_with_tool_reasoning.py" = ["E501"]
"tests/entrypoints/openai/test_completion_with_function_calling.py" = ["E501"]
"tests/entrypoints/openai/test_video.py" = ["E501"]
"tests/entrypoints/openai/test_vision.py" = ["E501"]
"tests/entrypoints/test_chat_utils.py" = ["E501"]
"tests/kernels/moe/modular_kernel_tools/common.py" = ["E501"]
"tests/models/language/generation/test_gemma.py" = ["E501"]
"tests/models/language/generation/test_mistral.py" = ["E501"]
"tests/models/multimodal/generation/test_ultravox.py" = ["E501"]
"tests/models/multimodal/generation/test_voxtral.py" = ["E501"]
"tests/models/multimodal/generation/vlm_utils/custom_inputs.py" = ["E501"]
"tests/tool_use/test_tool_choice_required.py" = ["E501"]
"tests/v1/attention/utils.py" = ["E501"]
"tests/v1/entrypoints/openai/responses/test_image.py" = ["E501"]
"tests/v1/kv_connector/nixl_integration/test_accuracy.py" = ["E501"]
"tests/v1/kv_connector/unit/test_offloading_connector.py" = ["E501"]
"tests/v1/logits_processors/test_custom_offline.py" = ["E501"]
"vllm/attention/ops/pallas_kv_cache_update.py" = ["E501"]
"vllm/compilation/collective_fusion.py" = ["E501"]
"vllm/compilation/wrapper.py" = ["E501"]
"vllm/config/vllm.py" = ["E501"]
"vllm/distributed/device_communicators/all2all.py" = ["E501"]
"vllm/entrypoints/openai/protocol.py" = ["E501"]
"vllm/lora/layers/vocal_parallel_embedding.py" = ["E501"]
"vllm/model_executor/model_loader/bitsandbytes_loader.py" = ["E501"]
"vllm/model_executor/models/bailing_moe.py" = ["E501"]
"vllm/model_executor/models/hyperclovax_vision.py" = ["E501"]
"vllm/model_executor/models/llama4_eagle.py" = ["E501"]
"vllm/model_executor/models/longcat_flash_mtp.py" = ["E501"]
"vllm/model_executor/models/phi4mm.py" = ["E501"]
"vllm/model_executor/models/qwen3_next.py" = ["E501"]
"vllm/model_executor/layers/quantization/ptpc_fp8.py" = ["E501"]
"vllm/v1/attention/backends/mla/common.py" = ["E501"]
"vllm/v1/engine/utils.py" = ["E501"]
"vllm/v1/utils.py" = ["E501"]
"vllm/v1/worker/gpu_model_runner.py" = ["E501"]
## Simplification rules
"tests/distributed/test_expert_placement.py" = ["SIM108"]
"tests/kernels/attention/test_cutlass_mla_decode.py" = ["SIM108"]
"tests/kernels/attention/test_flashmla.py" = ["SIM108"]
"tests/kernels/attention/test_lightning_attn.py" = ["SIM108"]
"tests/kernels/moe/test_pplx_moe.py" = ["SIM108"]
"tests/kernels/quantization/test_cutlass_scaled_mm.py" = ["SIM108"]
"tests/kernels/test_onednn.py" = ["SIM108"]
"tests/kernels/utils.py" = ["SIM108"]
"tests/multimodal/test_processing.py" = ["SIM108"]
"vllm/attention/ops/triton_reshape_and_cache_flash.py" = ["SIM108"]
"vllm/distributed/parallel_state.py" = ["SIM108"]
"vllm/entrypoints/chat_utils.py" = ["SIM108"]
"vllm/entrypoints/llm.py" = ["SIM108"]
"vllm/model_executor/layers/batch_invariant.py" = ["SIM108"]
"vllm/model_executor/layers/fla/ops/chunk_o.py" = ["SIM108"]
"vllm/model_executor/layers/fused_moe/fused_moe.py" = ["SIM108"]
"vllm/model_executor/layers/fused_moe/layer.py" = ["SIM108"]
"vllm/model_executor/layers/fused_moe/modular_kernel.py" = ["SIM108"]
"vllm/model_executor/layers/fused_moe/rocm_aiter_fused_moe.py" = ["SIM108"]
"vllm/model_executor/layers/layernorm.py" = ["SIM108"]
"vllm/model_executor/layers/lightning_attn.py" = ["SIM108"]
"vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors.py" = ["SIM103"]
"vllm/model_executor/layers/quantization/compressed_tensors/utils.py" = ["SIM110"]
"vllm/model_executor/layers/quantization/quark/utils.py" = ["SIM110"]
"vllm/utils/__init__.py" = ["SIM108"]
"vllm/v1/sample/ops/bad_words.py" = ["SIM108"]
"vllm/v1/sample/rejection_sampler.py" = ["SIM108"]
"vllm/v1/worker/tpu_model_runner.py" = ["SIM108"]
"vllm/_custom_ops.py" = ["SIM108"]
"tools/profiler/print_layerwise_table.py" = ["SIM118"]
## Loop variable binding issues
"tests/kernels/mamba/test_mamba_ssm_ssd.py" = ["B023"]
## Type annotation modernization and other rules
"vllm/attention/backends/abstract.py" = ["UP035", "UP006"]
"vllm/attention/layer.py" = ["UP035", "UP006"]
"vllm/attention/layers/chunked_local_attention.py" = ["UP035", "UP006"]
"vllm/attention/ops/flashmla.py" = ["UP035", "UP006"]
"vllm/attention/ops/paged_attn.py" = ["UP035", "UP006"]
"vllm/engine/arg_utils.py" = ["UP035", "UP006"]
"vllm/engine/metrics.py" = ["UP035", "UP006"]
"vllm/engine/metrics_types.py" = ["UP035", "UP006"]
"vllm/executor/executor_base.py" = ["UP035", "UP006"]
"vllm/executor/msgspec_utils.py" = ["UP035", "UP006"]
"vllm/executor/ray_distributed_executor.py" = ["UP035", "UP006", "SIM108", "SIM112"]
"vllm/executor/ray_utils.py" = ["UP035", "UP006"]
"vllm/executor/uniproc_executor.py" = ["UP035", "UP006"]
"vllm/model_executor/layers/fused_moe/flashinfer_trtllm_moe.py" = ["UP035"]
## Type comparison issues
"vllm/multimodal/inputs.py" = ["E721"]
# End of temporary ignores

[tool.ruff.lint]
select = [
    # pycodestyle
    "E",
    # Pyflakes
    "F",
    # pyupgrade
    "UP",
    # flake8-bugbear
    "B",
    # flake8-simplify
    "SIM",
    # isort
    "I",
    # flake8-logging-format
    "G",
]
ignore = [
    # star imports
    "F405", "F403",
    # lambda expression assignment
    "E731",
    # Loop control variable not used within loop body
    "B007",
    # f-string format
    "UP032",
    # Can remove once 3.10+ is the minimum Python version
    "UP007",
]

[tool.ruff.format]
docstring-code-format = true

[tool.mypy]
plugins = ['pydantic.mypy']
ignore_missing_imports = true
check_untyped_defs = true
follow_imports = "silent"

[tool.pytest.ini_options]
markers = [
    "slow_test",
    "skip_global_cleanup",
    "core_model: enable this model test in each PR instead of only nightly",
    "hybrid_model: models that contain mamba layers (including pure SSM and hybrid architectures)",
    "cpu_model: enable this model test in CPU tests",
    "cpu_test: mark test as CPU-only test",
    "split: run this test as part of a split",
    "distributed: run this test only in distributed GPU tests",
    "skip_v1: do not run this test with v1",
    "optional: optional tests that are automatically skipped, include --optional to run them",
]

[tool.ty.src]
root = "./vllm"
respect-ignore-files = true

[tool.ty.environment]
python = "./.venv"

[tool.typos.files]
# these files may be written in non english words
extend-exclude = ["tests/models/fixtures/*", "tests/prompts/*",
    "benchmarks/sonnet.txt", "tests/lora/data/*", "build/*",
    "vllm/third_party/*"]
ignore-hidden = true
ignore-files = true
ignore-dot = true
ignore-vcs = true
ignore-global = true
ignore-parent = true

[tool.typos.default]
binary = false
check-filename = false
check-file = true
unicode = true
ignore-hex = true
identifier-leading-digits = false
locale = "en"
extend-ignore-identifiers-re = ["NVML_*", ".*Unc.*", ".*_thw",
    ".*UE8M0.*", ".*[UE4M3|ue4m3].*", ".*eles.*",
     ".*[Tt]h[rR].*"]
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.default.extend-identifiers]
bbc5b7ede = "bbc5b7ede"
womens_doubles = "womens_doubles"
v_2nd = "v_2nd"
# splitted_input = "splitted_input"
NOOPs = "NOOPs"
typ = "typ"
nin_shortcut = "nin_shortcut"
UperNetDecoder = "UperNetDecoder"
subtile = "subtile"
cudaDevAttrMaxSharedMemoryPerBlockOptin = "cudaDevAttrMaxSharedMemoryPerBlockOptin"
SFOuput = "SFOuput"
# huggingface transformers repo uses these words
depthwise_seperable_out_channel = "depthwise_seperable_out_channel"
DepthWiseSeperableConv1d = "DepthWiseSeperableConv1d"
depthwise_seperable_CNN = "depthwise_seperable_CNN"

[tool.typos.default.extend-words]
iy = "iy"
tendencias = "tendencias"
# intel cpu features
tme = "tme"
dout = "dout"
Pn = "Pn"
arange = "arange"

[tool.typos.type.py]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.py.extend-identifiers]
arange = "arange"
NDArray = "NDArray"
EOFError = "EOFError"
fo = "fo"
ba = "ba"

[tool.typos.type.py.extend-words]
ba = "ba"

[tool.typos.type.cpp]
extend-glob = ["*.cu"]
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.cpp.extend-identifiers]
countr_one = "countr_one"
k_ot = "k_ot"
ot = "ot"

[tool.typos.type.cpp.extend-words]

[tool.typos.type.rust]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.rust.extend-identifiers]
flate2 = "flate2"

[tool.typos.type.rust.extend-words]
ser = "ser"

[tool.typos.type.lock]
extend-glob = []
check-file = false
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.lock.extend-identifiers]

[tool.typos.type.lock.extend-words]

[tool.typos.type.jl]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.jl.extend-identifiers]

[tool.typos.type.jl.extend-words]
modul = "modul"
egals = "egals"
usig = "usig"
egal = "egal"

[tool.typos.type.go]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.go.extend-identifiers]
flate = "flate"

[tool.typos.type.go.extend-words]

[tool.typos.type.css]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.css.extend-identifiers]
nd = "nd"

[tool.typos.type.css.extend-words]

[tool.typos.type.man]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.man.extend-identifiers]
Nd = "Nd"

[tool.typos.type.man.extend-words]

[tool.typos.type.cert]
extend-glob = []
check-file = false
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.cert.extend-identifiers]

[tool.typos.type.cert.extend-words]

[tool.typos.type.sh]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.sh.extend-identifiers]
ot = "ot"

[tool.typos.type.sh.extend-words]

[tool.typos.type.vimscript]
extend-glob = []
extend-ignore-identifiers-re = []
extend-ignore-words-re = []
extend-ignore-re = []

[tool.typos.type.vimscript.extend-identifiers]
windo = "windo"

[tool.typos.type.vimscript.extend-words]

[tool.uv]
no-build-isolation-package = ["torch"]