usage: vllm serve [model_tag] [options]

Launch a local OpenAI-compatible API server to serve LLM
completions via HTTP. Defaults to Qwen/Qwen3-0.6B if no model is specified.

Search by using: `--help=<ConfigGroup>` to explore options by section (e.g.,
--help=ModelConfig, --help=Frontend)
  Use `--help=all` to show all available flags at once.

positional arguments:
  model_tag             The model tag to serve (optional if specified in
                        config) (default: None)

options:
  --aggregate-engine-logging
                        Log aggregate rather than per-engine statistics when
                        using data parallelism. (default: False)
  --api-server-count API_SERVER_COUNT, -asc API_SERVER_COUNT
                        How many API server processes to run. Defaults to
                        data_parallel_size if not specified. (default: None)
  --config CONFIG       Read CLI options from a config file. Must be a YAML
                        with the following options: https://docs.vllm.ai/en/la
                        test/configuration/serve_args.html (default: None)
  --disable-log-stats   Disable logging statistics. (default: False)
  --enable-log-requests, --no-enable-log-requests
                        Enable logging request information, dependent on log
                        level: - INFO: Request ID, parameters and LoRA
                        request. - DEBUG: Prompt inputs (e.g: text, token
                        IDs). You can set the minimum log level via
                        `VLLM_LOGGING_LEVEL`. (default: False)
  --fail-on-environ-validation, --no-fail-on-environ-validation
                        If set, the engine will raise an error if environment
                        validation fails. (default: False)
  --grpc                Launch a gRPC server instead of the HTTP OpenAI-
                        compatible server. Requires: pip install vllm[grpc].
                        (default: False)
  --headless            Run in headless mode. See multi-node data parallel
                        documentation for more details. (default: False)
  -h, --help            show this help message and exit

Frontend:
  Arguments for the OpenAI-compatible frontend server.

  --allow-credentials, --no-allow-credentials
                        Allow credentials. (default: False)
  --allowed-headers ALLOWED_HEADERS
                        Allowed headers. (default: ['*'])
  --allowed-methods ALLOWED_METHODS
                        Allowed methods. (default: ['*'])
  --allowed-origins ALLOWED_ORIGINS
                        Allowed origins. (default: ['*'])
  --api-key API_KEY [API_KEY ...]
                        If provided, the server will require one of these keys
                        to be presented in the header. (default: None)
  --chat-template CHAT_TEMPLATE
  --chat-template-content-format {auto,openai,string}
  --default-chat-template-kwargs DEFAULT_CHAT_TEMPLATE_KWARGS
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --disable-access-log-for-endpoints DISABLE_ACCESS_LOG_FOR_ENDPOINTS
                        Comma-separated list of endpoint paths to exclude from
                        uvicorn access logs. This is useful to reduce log
                        noise from high-frequency endpoints like health
                        checks. Example: "/health,/metrics,/ping". When set,
                        access logs for requests to these paths will be
                        suppressed while keeping logs for other endpoints.
                        (default: None)
  --disable-fastapi-docs, --no-disable-fastapi-docs
                        Disable FastAPI's OpenAPI schema, Swagger UI, and
                        ReDoc endpoint. (default: False)
  --disable-frontend-multiprocessing, --no-disable-frontend-multiprocessing
  --disable-uvicorn-access-log, --no-disable-uvicorn-access-log
                        Disable uvicorn access log. (default: False)
  --enable-auto-tool-choice, --no-enable-auto-tool-choice
  --enable-force-include-usage, --no-enable-force-include-usage
  --enable-log-deltas, --no-enable-log-deltas
  --enable-log-outputs, --no-enable-log-outputs
  --enable-offline-docs, --no-enable-offline-docs
                        Enable offline FastAPI documentation for air-gapped
                        environments. Uses vendored static assets bundled with
                        vLLM. (default: False)
  --enable-prompt-tokens-details, --no-enable-prompt-tokens-details
  --enable-request-id-headers, --no-enable-request-id-headers
                        If specified, API server will add X-Request-Id header
                        to responses. (default: False)
  --enable-server-load-tracking, --no-enable-server-load-tracking
  --enable-ssl-refresh, --no-enable-ssl-refresh
                        Refresh SSL Context when SSL certificate files change
                        (default: False)
  --enable-tokenizer-info-endpoint, --no-enable-tokenizer-info-endpoint
  --exclude-tools-when-tool-choice-none, --no-exclude-tools-when-tool-choice-none
  --h11-max-header-count H11_MAX_HEADER_COUNT
                        Maximum number of HTTP headers allowed in a request
                        for h11 parser. Helps mitigate header abuse. Default:
                        256. (default: 256)
  --h11-max-incomplete-event-size H11_MAX_INCOMPLETE_EVENT_SIZE
                        Maximum size (bytes) of an incomplete HTTP event
                        (header or body) for h11 parser. Helps mitigate header
                        abuse. Default: 4194304 (4 MB). (default: 4194304)
  --host HOST           Host name. (default: None)
  --log-config-file LOG_CONFIG_FILE
  --log-error-stack, --no-log-error-stack
  --lora-modules LORA_MODULES [LORA_MODULES ...]
  --max-log-len MAX_LOG_LEN
  --middleware MIDDLEWARE
                        Additional ASGI middleware to apply to the app. We
                        accept multiple --middleware arguments. The value
                        should be an import path. If a function is provided,
                        vLLM will add it to the server using
                        `@app.middleware('http')`. If a class is provided,
                        vLLM will add it to the server using
                        `app.add_middleware()`. (default: [])
  --port PORT           Port number. (default: 8000)
  --response-role RESPONSE_ROLE
  --return-tokens-as-token-ids, --no-return-tokens-as-token-ids
  --root-path ROOT_PATH
                        FastAPI root_path when app is behind a path based
                        routing proxy. (default: None)
  --ssl-ca-certs SSL_CA_CERTS
                        The CA certificates file. (default: None)
  --ssl-cert-reqs SSL_CERT_REQS
                        Whether client certificate is required (see stdlib ssl
                        module's). (default: 0)
  --ssl-certfile SSL_CERTFILE
                        The file path to the SSL cert file. (default: None)
  --ssl-ciphers SSL_CIPHERS
                        SSL cipher suites for HTTPS (TLS 1.2 and below only).
                        Example: 'ECDHE-RSA-AES256-GCM-SHA384:ECDHE-RSA-
                        CHACHA20-POLY1305' (default: None)
  --ssl-keyfile SSL_KEYFILE
                        The file path to the SSL key file. (default: None)
  --tokens-only, --no-tokens-only
  --tool-call-parser {deepseek_v3,deepseek_v31,deepseek_v32,ernie45,functiongemma,gigachat3,glm45,glm47,granite,granite-20b-fc,hermes,hunyuan_a13b,internlm,jamba,kimi_k2,llama3_json,llama4_json,llama4_pythonic,longcat,minimax,minimax_m2,mistral,olmo3,openai,phi4_mini_json,pythonic,qwen3_coder,qwen3_xml,seed_oss,step3,step3p5,xlam} or name registered in --tool-parser-plugin
  --tool-parser-plugin TOOL_PARSER_PLUGIN
  --tool-server TOOL_SERVER
  --trust-request-chat-template, --no-trust-request-chat-template
  --uds UDS             Unix domain socket path. If set, host and port
                        arguments are ignored. (default: None)
  --uvicorn-log-level {critical,debug,error,info,trace,warning}
                        Log level for uvicorn. (default: info)

ModelConfig:
  Configuration for the model.

  --allow-deprecated-quantization, --no-allow-deprecated-quantization
                        Whether to allow deprecated quantization methods.
                        (default: False)
  --allowed-local-media-path ALLOWED_LOCAL_MEDIA_PATH
                        Allowing API requests to read local images or videos
                        from directories specified by the server file system.
                        This is a security risk. Should only be enabled in
                        trusted environments. (default: )
  --allowed-media-domains ALLOWED_MEDIA_DOMAINS [ALLOWED_MEDIA_DOMAINS ...]
                        If set, only media URLs that belong to this domain can
                        be used for multi-modal inputs. (default: None)
  --code-revision CODE_REVISION
                        The specific revision to use for the model code on the
                        Hugging Face Hub. It can be a branch name, a tag name,
                        or a commit id. If unspecified, will use the default
                        version. (default: None)
  --config-format ['auto', 'hf', 'mistral']
                        The format of the model config to load:
                        - "auto" will try to load the config in hf format if
                        available after trying to load in mistral format.
                        - "hf" will load the config in hf format.
                        - "mistral" will load the config in mistral format.
                        (default: auto)
  --convert {auto,classify,embed,none}
                        Convert the model using adapters defined in
                        [vllm.model_executor.models.adapters][]. The most
                        common use case is to adapt a text generation model to
                        be used for pooling tasks. (default: auto)
  --disable-cascade-attn, --no-disable-cascade-attn
                        Disable cascade attention for V1. While cascade
                        attention does not change the mathematical
                        correctness, disabling it could be useful for
                        preventing potential numerical issues. Note that even
                        if this is set to False, cascade attention will be
                        only used when the heuristic tells that it's
                        beneficial. (default: False)
  --disable-sliding-window, --no-disable-sliding-window
                        Whether to disable sliding window. If True, we will
                        disable the sliding window functionality of the model,
                        capping to sliding window size. If the model does not
                        support sliding window, this argument is ignored.
                        (default: False)
  --dtype {auto,bfloat16,float,float16,float32,half}
                        Data type for model weights and activations:
                        - "auto" will use FP16 precision for FP32 and FP16
                        models, and BF16 precision for BF16 models.
                        - "half" for FP16. Recommended for AWQ quantization.
                        - "float16" is the same as "half".
                        - "bfloat16" for a balance between precision and
                        range.
                        - "float" is shorthand for FP32 precision.
                        - "float32" for FP32 precision. (default: auto)
  --enable-prompt-embeds, --no-enable-prompt-embeds
                        If `True`, enables passing text embeddings as inputs
                        via the `prompt_embeds` key.
                        WARNING: The vLLM engine may crash if incorrect shape
                        of embeddings is passed. Only enable this flag for
                        trusted users! (default: False)
  --enable-return-routed-experts, --no-enable-return-routed-experts
                        Whether to return routed experts. (default: False)
  --enable-sleep-mode, --no-enable-sleep-mode
                        Enable sleep mode for the engine (only cuda and hip
                        platforms are supported). (default: False)
  --enforce-eager, --no-enforce-eager
                        Whether to always use eager-mode PyTorch. If True, we
                        will disable CUDA graph and always execute the model
                        in eager mode. If False, we will use CUDA graph and
                        eager execution in hybrid for maximal performance and
                        flexibility. (default: False)
  --generation-config GENERATION_CONFIG
                        The folder path to the generation config. Defaults to
                        `"auto"`, the generation config will be loaded from
                        model path. If set to `"vllm"`, no generation config
                        is loaded, vLLM defaults will be used. If set to a
                        folder path, the generation config will be loaded from
                        the specified folder path. If `max_new_tokens` is
                        specified in generation config, then it sets a server-
                        wide limit on the number of output tokens for all
                        requests. (default: auto)
  --hf-config-path HF_CONFIG_PATH
                        Name or path of the Hugging Face config to use. If
                        unspecified, model name or path will be used.
                        (default: None)
  --hf-overrides HF_OVERRIDES
                        If a dictionary, contains arguments to be forwarded to
                        the Hugging Face config. If a callable, it is called
                        to update the HuggingFace config. (default: {})
  --hf-token [HF_TOKEN]
                        The token to use as HTTP bearer authorization for
                        remote files . If `True`, will use the token generated
                        when running `hf auth login` (stored in
                        `~/.cache/huggingface/token`). (default: None)
  --io-processor-plugin IO_PROCESSOR_PLUGIN
                        IOProcessor plugin name to load at model startup
                        (default: None)
  --logits-processors LOGITS_PROCESSORS [LOGITS_PROCESSORS ...]
                        One or more logits processors' fully-qualified class
                        names or class definitions (default: None)
  --logprobs-mode {processed_logits,processed_logprobs,raw_logits,raw_logprobs}
                        Indicates the content returned in the logprobs and
                        prompt_logprobs. Supported mode: 1) raw_logprobs, 2)
                        processed_logprobs, 3) raw_logits, 4)
                        processed_logits. Raw means the values before applying
                        any logit processors, like bad words. Processed means
                        the values after applying all processors, including
                        temperature and top_k/top_p. (default: raw_logprobs)
  --max-logprobs MAX_LOGPROBS
                        Maximum number of log probabilities to return when
                        `logprobs` is specified in `SamplingParams`. The
                        default value comes the default for the OpenAI Chat
                        Completions API. -1 means no cap, i.e. all
                        (output_length * vocab_size) logprobs are allowed to
                        be returned and it may cause OOM. (default: 20)
  --max-model-len MAX_MODEL_LEN
                        Model context length (prompt and output). If
                        unspecified, will be automatically derived from the
                        model config.
                        When passing via `--max-model-len`, supports
                        k/m/g/K/M/G in human-readable format. Examples:
                        - 1k -> 1000
                        - 1K -> 1024
                        - 25.6k -> 25,600
                        - -1 or 'auto' -> Automatically choose the maximum
                        model length that fits in GPU memory. This will use
                        the model's maximum context length if it fits,
                        otherwise it will find the largest length that can be
                        accommodated.
                        Parse human-readable integers like '1k', '2M', etc.
                        Including decimal values with decimal multipliers.
                        Also accepts -1 or 'auto' as a special value for auto-
                        detection.
                        Examples: - '1k' -> 1,000 - '1K' -> 1,024 - '25.6k' ->
                        25,600 - '-1' or 'auto' -> -1 (special value for auto-
                        detection) (default: None)
  --model MODEL         Name or path of the Hugging Face model to use. It is
                        also used as the content for `model_name` tag in
                        metrics output when `served_model_name` is not
                        specified. (default: Qwen/Qwen3-0.6B)
  --model-impl ['auto', 'terratorch', 'transformers', 'vllm']
                        Which implementation of the model to use:
                        - "auto" will try to use the vLLM implementation, if
                        it exists, and fall back to the Transformers
                        implementation if no vLLM implementation is available.
                        - "vllm" will use the vLLM model implementation.
                        - "transformers" will use the Transformers model
                        implementation.
                        - "terratorch" will use the TerraTorch model
                        implementation. (default: auto)
  --override-attention-dtype OVERRIDE_ATTENTION_DTYPE
                        Override dtype for attention (default: None)
  --override-generation-config OVERRIDE_GENERATION_CONFIG
                        Overrides or sets generation config. e.g.
                        `{"temperature": 0.5}`. If used with `--generation-
                        config auto`, the override parameters will be merged
                        with the default config from the model. If used with
                        `--generation-config vllm`, only the override
                        parameters are used.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: {})
  --pooler-config POOLER_CONFIG
                        Pooler config which controls the behaviour of output
                        pooling in pooling models.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --quantization QUANTIZATION, -q QUANTIZATION
                        Method used to quantize the weights. If `None`, we
                        first check the `quantization_config` attribute in the
                        model config file. If that is `None`, we assume the
                        model weights are not quantized and use `dtype` to
                        determine the data type of the weights. (default:
                        None)
  --revision REVISION   The specific model version to use. It can be a branch
                        name, a tag name, or a commit id. If unspecified, will
                        use the default version. (default: None)
  --runner {auto,draft,generate,pooling}
                        The type of model runner to use. Each vLLM instance
                        only supports one model runner, even if the same model
                        can be used for multiple types. (default: auto)
  --seed SEED           Random seed for reproducibility.
                        We must set the global seed because otherwise,
                        different tensor parallel workers would sample
                        different tokens, leading to inconsistent results.
                        (default: 0)
  --served-model-name SERVED_MODEL_NAME [SERVED_MODEL_NAME ...]
                        The model name(s) used in the API. If multiple names
                        are provided, the server will respond to any of the
                        provided names. The model name in the model field of a
                        response will be the first name in this list. If not
                        specified, the model name will be the same as the
                        `--model` argument. Noted that this name(s) will also
                        be used in `model_name` tag content of prometheus
                        metrics, if multiple names provided, metrics tag will
                        take the first one. (default: None)
  --skip-tokenizer-init, --no-skip-tokenizer-init
                        Skip initialization of tokenizer and detokenizer.
                        Expects valid `prompt_token_ids` and `None` for prompt
                        from the input. The generated output will contain
                        token ids. (default: False)
  --tokenizer TOKENIZER
                        Name or path of the Hugging Face tokenizer to use. If
                        unspecified, model name or path will be used.
                        (default: None)
  --tokenizer-mode ['auto', 'deepseek_v32', 'hf', 'mistral', 'slow']
                        Tokenizer mode:
                        - "auto" will use the tokenizer from `mistral_common`
                        for Mistral models if available, otherwise it will use
                        the "hf" tokenizer.
                        - "hf" will use the fast tokenizer if available.
                        - "slow" will always use the slow tokenizer.
                        - "mistral" will always use the tokenizer from
                        `mistral_common`.
                        - "deepseek_v32" will always use the tokenizer from
                        `deepseek_v32`.
                        - "qwen_vl" will always use the tokenizer from
                        `qwen_vl`.
                        - Other custom values can be supported via plugins.
                        (default: auto)
  --tokenizer-revision TOKENIZER_REVISION
                        The specific revision to use for the tokenizer on the
                        Hugging Face Hub. It can be a branch name, a tag name,
                        or a commit id. If unspecified, will use the default
                        version. (default: None)
  --trust-remote-code, --no-trust-remote-code
                        Trust remote code (e.g., from HuggingFace) when
                        downloading the model and tokenizer. (default: False)

LoadConfig:
  Configuration for loading the model weights.

  --download-dir DOWNLOAD_DIR
                        Directory to download and load the weights, default to
                        the default cache directory of Hugging Face. (default:
                        None)
  --ignore-patterns IGNORE_PATTERNS [IGNORE_PATTERNS ...]
                        The list of patterns to ignore when loading the model.
                        Default to "original/**/*" to avoid repeated loading
                        of llama's checkpoints. (default: ['original/**/*'])
  --load-format LOAD_FORMAT
                        The format of the model weights to load:
                        - "auto" will try to load the weights in the
                        safetensors format and fall back to the pytorch bin
                        format if safetensors format is not available.
                        - "pt" will load the weights in the pytorch bin
                        format.
                        - "safetensors" will load the weights in the
                        safetensors format.
                        - "npcache" will load the weights in pytorch format
                        and store a numpy cache to speed up the loading.
                        - "dummy" will initialize the weights with random
                        values, which is mainly for profiling.
                        - "tensorizer" will use CoreWeave's tensorizer library
                        for fast weight loading. See the Tensorize vLLM Model
                        script in the Examples section for more information.
                        - "runai_streamer" will load the Safetensors weights
                        using Run:ai Model Streamer.
                        - "runai_streamer_sharded" will load weights from pre-
                        sharded checkpoint files using Run:ai Model Streamer.
                        - "bitsandbytes" will load the weights using
                        bitsandbytes quantization.
                        - "sharded_state" will load weights from pre-sharded
                        checkpoint files, supporting efficient loading of
                        tensor-parallel models.
                        - "gguf" will load weights from GGUF format files
                        (details specified in https://github.com/ggml-
                        org/ggml/blob/master/docs/gguf.md).
                        - "mistral" will load weights from consolidated
                        safetensors files used by Mistral models. - Other
                        custom values can be supported via plugins. (default:
                        auto)
  --model-loader-extra-config MODEL_LOADER_EXTRA_CONFIG
                        Extra config for model loader. This will be passed to
                        the model loader corresponding to the chosen
                        load_format. (default: {})
  --pt-load-map-location PT_LOAD_MAP_LOCATION
                        pt_load_map_location: the map location for loading
                        pytorch checkpoint, to support loading checkpoints can
                        only be loaded on certain devices like "cuda", this is
                        equivalent to {"": "cuda"}. Another supported format
                        is mapping from different devices like from GPU 1 to
                        GPU 0: {"cuda:1": "cuda:0"}. Note that when passed
                        from command line, the strings in dictionary needs to
                        be double quoted for json parsing. For more details,
                        see original doc for `map_location` in https://pytorch
                        .org/docs/stable/generated/torch.load.html (default:
                        cpu)
  --safetensors-load-strategy SAFETENSORS_LOAD_STRATEGY
                        Specifies the loading strategy for safetensors
                        weights. - "lazy" (default): Weights are memory-mapped
                        from the file. This enables on-demand loading and is
                        highly efficient for models on local storage. -
                        "eager": The entire file is read into CPU memory
                        upfront before loading. This is recommended for models
                        on network filesystems (e.g., Lustre, NFS) as it
                        avoids inefficient random reads, significantly
                        speeding up model initialization. However, it uses
                        more CPU RAM. - "torchao": Weights are loaded in
                        upfront and then reconstructed into torchao tensor
                        subclasses. This is used when the checkpoint was
                        quantized using torchao and saved using safetensors.
                        Needs torchao >= 0.14.0 (default: lazy)
  --use-tqdm-on-load, --no-use-tqdm-on-load
                        Whether to enable tqdm for showing progress bar when
                        loading model weights. (default: True)

AttentionConfig:
  Configuration for attention mechanisms in vLLM.

  --attention-backend ATTENTION_BACKEND
                        Attention backend to use. Use "auto" or None for
                        automatic selection. (default: None)

StructuredOutputsConfig:
  Dataclass which contains structured outputs config for the engine.

  --reasoning-parser REASONING_PARSER
                        Select the reasoning parser depending on the model
                        that you're using. This is used to parse the reasoning
                        content into OpenAI API format. (default: )
  --reasoning-parser-plugin REASONING_PARSER_PLUGIN
                        Path to a dynamically reasoning parser plugin that can
                        be dynamically loaded and registered. (default: )

ParallelConfig:
  Configuration for the distributed execution.

  --all2all-backend {allgather_reducescatter,deepep_high_throughput,deepep_low_latency,flashinfer_all2allv,mori,naive,pplx}
                        All2All backend for MoE expert parallel communication.
                        Available options:
                        - "naive": Naive all2all implementation using
                        broadcasts
                        - "allgather_reducescatter": All2all based on
                        allgather and reducescatter
                        - "deepep_high_throughput": Use deepep high-throughput
                        kernels
                        - "deepep_low_latency": Use deepep low-latency kernels
                        - "mori": Use mori kernels
                        - "flashinfer_all2allv": Use flashinfer alltoallv
                        kernels for mnnvl (default: allgather_reducescatter)
  --cp-kv-cache-interleave-size CP_KV_CACHE_INTERLEAVE_SIZE
                        Interleave size of kv_cache storage while using DCP or
                        PCP. For `total_cp_rank = pcp_rank * dcp_world_size +
                        dcp_rank`, and `total_cp_world_size = pcp_world_size *
                        dcp_world_size`. store interleave_size tokens on
                        total_cp_rank i, then store next interleave_size
                        tokens on total_cp_rank i+1. Interleave_size=1: token-
                        level alignment, where token `i` is stored on
                        total_cp_rank `i % total_cp_world_size`.
                        Interleave_size=block_size: block-level alignment,
                        where tokens are first populated to the preceding
                        ranks. Tokens are then stored in (rank i+1, block j)
                        only after (rank i, block j) is fully occupied.
                        Block_size should be greater than or equal to
                        cp_kv_cache_interleave_size. Block_size should be
                        divisible by cp_kv_cache_interleave_size. (default: 1)
  --data-parallel-address DATA_PARALLEL_ADDRESS, -dpa DATA_PARALLEL_ADDRESS
                        Address of data parallel cluster head-node. (default:
                        None)
  --data-parallel-backend DATA_PARALLEL_BACKEND, -dpb DATA_PARALLEL_BACKEND
                        Backend for data parallel, either "mp" or "ray".
                        (default: mp)
  --data-parallel-external-lb, --no-data-parallel-external-lb, -dpe
                        Whether to use "external" DP LB mode. Applies only to
                        online serving and when data_parallel_size > 0. This
                        is useful for a "one-pod-per-rank" wide-EP setup in
                        Kubernetes. Set implicitly when --data-parallel-rank
                        is provided explicitly to vllm serve. (default: False)
  --data-parallel-hybrid-lb, --no-data-parallel-hybrid-lb, -dph
                        Whether to use "hybrid" DP LB mode. Applies only to
                        online serving and when data_parallel_size > 0.
                        Enables running an AsyncLLM and API server on a "per-
                        node" basis where vLLM load balances between local
                        data parallel ranks, but an external LB balances
                        between vLLM nodes/replicas. Set explicitly in
                        conjunction with --data-parallel-start-rank. (default:
                        False)
  --data-parallel-rank DATA_PARALLEL_RANK, -dpn DATA_PARALLEL_RANK
                        Data parallel rank of this instance. When set, enables
                        external load balancer mode. (default: None)
  --data-parallel-rpc-port DATA_PARALLEL_RPC_PORT, -dpp DATA_PARALLEL_RPC_PORT
                        Port for data parallel RPC communication. (default:
                        None)
  --data-parallel-size DATA_PARALLEL_SIZE, -dp DATA_PARALLEL_SIZE
                        Number of data parallel groups. MoE layers will be
                        sharded according to the product of the tensor
                        parallel size and data parallel size. (default: 1)
  --data-parallel-size-local DATA_PARALLEL_SIZE_LOCAL, -dpl DATA_PARALLEL_SIZE_LOCAL
                        Number of data parallel replicas to run on this node.
                        (default: None)
  --data-parallel-start-rank DATA_PARALLEL_START_RANK, -dpr DATA_PARALLEL_START_RANK
                        Starting data parallel rank for secondary nodes.
                        (default: None)
  --dbo-decode-token-threshold DBO_DECODE_TOKEN_THRESHOLD
                        The threshold for dual batch overlap for batches only
                        containing decodes. If the number of tokens in the
                        request is greater than this threshold, microbatching
                        will be used. Otherwise, the request will be processed
                        in a single batch. (default: 32)
  --dbo-prefill-token-threshold DBO_PREFILL_TOKEN_THRESHOLD
                        The threshold for dual batch overlap for batches that
                        contain one or more prefills. If the number of tokens
                        in the request is greater than this threshold,
                        microbatching will be used. Otherwise, the request
                        will be processed in a single batch. (default: 512)
  --dcp-comm-backend {a2a,ag_rs}
                        Communication backend for Decode Context Parallel
                        (DCP). - "ag_rs": AllGather + ReduceScatter (default,
                        existing behavior) - "a2a": All-to-All exchange of
                        partial outputs + LSE, then combine with Triton
                        kernel. Reduces NCCL calls from 3 to 2 per layer for
                        MLA models. (default: ag_rs)
  --dcp-kv-cache-interleave-size DCP_KV_CACHE_INTERLEAVE_SIZE
                        Interleave size of kv_cache storage while using DCP.
                        dcp_kv_cache_interleave_size has been replaced by
                        cp_kv_cache_interleave_size, and will be deprecated
                        when PCP is fully supported. (default: 1)
  --decode-context-parallel-size DECODE_CONTEXT_PARALLEL_SIZE, -dcp DECODE_CONTEXT_PARALLEL_SIZE
                        Number of decode context parallel groups, because the
                        world size does not change by dcp, it simply reuse the
                        GPUs of TP group, and tp_size needs to be divisible by
                        dcp_size. (default: 1)
  --disable-custom-all-reduce, --no-disable-custom-all-reduce
                        Disable the custom all-reduce kernel and fall back to
                        NCCL. (default: False)
  --disable-nccl-for-dp-synchronization, --no-disable-nccl-for-dp-synchronization
                        Forces the dp synchronization logic in
                        vllm/v1/worker/dp_utils.py  to use Gloo instead of
                        NCCL for its all reduce.
                        Defaults to True when async scheduling is enabled,
                        False otherwise. (default: None)
  --distributed-executor-backend ['external_launcher', 'mp', 'ray', 'uni']
                        Backend to use for distributed model workers, either
                        "ray" or "mp" (multiprocessing). If the product of
                        pipeline_parallel_size and tensor_parallel_size is
                        less than or equal to the number of GPUs available,
                        "mp" will be used to keep processing on a single host.
                        Otherwise, an error will be raised. To use "mp" you
                        must also set nnodes, and to use "ray" you must
                        manually set distributed_executor_backend to "ray".
                        Note that tpu only support Ray for distributed
                        inference. (default: None)
  --distributed-timeout-seconds DISTRIBUTED_TIMEOUT_SECONDS
                        Timeout in seconds for distributed operations (e.g.,
                        init_process_group). If set, this value is passed to
                        torch.distributed.init_process_group as the timeout
                        parameter. If None, PyTorch's default timeout is used
                        (600s for NCCL). Increase this for multi-node setups
                        where model downloads may be slow. (default: None)
  --enable-dbo, --no-enable-dbo
                        Enable dual batch overlap for the model executor.
                        (default: False)
  --enable-elastic-ep, --no-enable-elastic-ep
                        Enable elastic expert parallelism with stateless NCCL
                        groups for DP/EP. (default: False)
  --enable-eplb, --no-enable-eplb
                        Enable expert parallelism load balancing for MoE
                        layers. (default: False)
  --enable-expert-parallel, --no-enable-expert-parallel, -ep
                        Use expert parallelism instead of tensor parallelism
                        for MoE layers. (default: False)
  --eplb-config EPLB_CONFIG
                        Expert parallelism configuration.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default:
                        EPLBConfig(window_size=1000, step_interval=3000,
                        num_redundant_experts=0, log_balancedness=False,
                        log_balancedness_interval=1, use_async=False,
                        policy='default'))
  --expert-placement-strategy {linear,round_robin}
                        The expert placement strategy for MoE layers:
                        - "linear": Experts are placed in a contiguous manner.
                        For example, with 4 experts and 2 ranks, rank 0 will
                        have experts [0, 1] and rank 1 will have experts [2,
                        3].
                        - "round_robin": Experts are placed in a round-robin
                        manner. For example, with 4 experts and 2 ranks, rank
                        0 will have experts [0, 2] and rank 1 will have
                        experts [1, 3]. This strategy can help improve load
                        balancing for grouped expert models with no redundant
                        experts. (default: linear)
  --master-addr MASTER_ADDR
                        distributed master address for multi-node distributed
                        inference when distributed_executor_backend is mp.
                        (default: 127.0.0.1)
  --master-port MASTER_PORT
                        distributed master port for multi-node distributed
                        inference when distributed_executor_backend is mp.
                        (default: 29501)
  --max-parallel-loading-workers MAX_PARALLEL_LOADING_WORKERS
                        Maximum number of parallel loading workers when
                        loading model sequentially in multiple batches. To
                        avoid RAM OOM when using tensor parallel and large
                        models. (default: None)
  --nnodes NNODES, -n NNODES
                        num of nodes for multi-node distributed inference when
                        distributed_executor_backend is mp. (default: 1)
  --node-rank NODE_RANK, -r NODE_RANK
                        distributed node rank for multi-node distributed
                        inference when distributed_executor_backend is mp.
                        (default: 0)
  --pipeline-parallel-size PIPELINE_PARALLEL_SIZE, -pp PIPELINE_PARALLEL_SIZE
                        Number of pipeline parallel groups. (default: 1)
  --prefill-context-parallel-size PREFILL_CONTEXT_PARALLEL_SIZE, -pcp PREFILL_CONTEXT_PARALLEL_SIZE
                        Number of prefill context parallel groups. (default:
                        1)
  --ray-workers-use-nsight, --no-ray-workers-use-nsight
                        Whether to profile Ray workers with nsight, see
                        https://docs.ray.io/en/latest/ray-observability/user-
                        guides/profiling.html#profiling-nsight-profiler.
                        (default: False)
  --tensor-parallel-size TENSOR_PARALLEL_SIZE, -tp TENSOR_PARALLEL_SIZE
                        Number of tensor parallel groups. (default: 1)
  --ubatch-size UBATCH_SIZE
                        Number of ubatch size. (default: 0)
  --worker-cls WORKER_CLS
                        The full name of the worker class to use. If "auto",
                        the worker class will be determined based on the
                        platform. (default: auto)
  --worker-extension-cls WORKER_EXTENSION_CLS
                        The full name of the worker extension class to use.
                        The worker extension class is dynamically inherited by
                        the worker class. This is used to inject new
                        attributes and methods to the worker class for use in
                        collective_rpc calls. (default: )

CacheConfig:
  Configuration for the KV cache.

  --block-size BLOCK_SIZE
                        Size of a contiguous cache block in number of tokens.
                        Accepts None (meaning "use default"). After
                        construction, always int. (default: None)
  --calculate-kv-scales, --no-calculate-kv-scales
                        This enables dynamic calculation of `k_scale` and
                        `v_scale` when kv_cache_dtype is fp8. If `False`, the
                        scales will be loaded from the model checkpoint if
                        available. Otherwise, the scales will default to 1.0.
                        (default: False)
  --enable-prefix-caching, --no-enable-prefix-caching
                        Whether to enable prefix caching. (default: None)
  --gpu-memory-utilization GPU_MEMORY_UTILIZATION
                        The fraction of GPU memory to be used for the model
                        executor, which can range from 0 to 1. For example, a
                        value of 0.5 would imply 50% GPU memory utilization.
                        If unspecified, will use the default value of 0.9.
                        This is a per-instance limit, and only applies to the
                        current vLLM instance. It does not matter if you have
                        another vLLM instance running on the same GPU. For
                        example, if you have two vLLM instances running on the
                        same GPU, you can set the GPU memory utilization to
                        0.5 for each instance. (default: 0.9)
  --kv-cache-dtype {auto,bfloat16,fp8,fp8_ds_mla,fp8_e4m3,fp8_e5m2,fp8_inc,half}
                        Data type for kv cache storage. If "auto", will use
                        model data type. CUDA 11.8+ supports fp8 (=fp8_e4m3)
                        and fp8_e5m2. ROCm (AMD GPU) supports fp8 (=fp8_e4m3).
                        Intel Gaudi (HPU) supports fp8 (using fp8_inc). Some
                        models (namely DeepSeekV3.2) default to fp8, set to
                        bfloat16 to use bfloat16 instead, this is an invalid
                        option for models that do not default to fp8.
                        (default: auto)
  --kv-cache-memory-bytes KV_CACHE_MEMORY_BYTES
                        Size of KV Cache per GPU in bytes. By default, this is
                        set to None and vllm can automatically infer the kv
                        cache size based on gpu_memory_utilization. However,
                        users may want to manually specify the kv cache memory
                        size. kv_cache_memory_bytes allows more fine-grain
                        control of how much memory gets used when compared
                        with using gpu_memory_utilization. Note that
                        kv_cache_memory_bytes (when not-None) ignores
                        gpu_memory_utilization
                        Parse human-readable integers like '1k', '2M', etc.
                        Including decimal values with decimal multipliers.
                        Examples: - '1k' -> 1,000 - '1K' -> 1,024 - '25.6k' ->
                        25,600 (default: None)
  --kv-offloading-backend {lmcache,native}
                        The backend to use for KV cache offloading. Supported
                        backends include 'native' (vLLM native CPU
                        offloading), 'lmcache'. KV offloading is only
                        activated when kv_offloading_size is set. (default:
                        native)
  --kv-offloading-size KV_OFFLOADING_SIZE
                        Size of the KV cache offloading buffer in GiB. When TP
                        > 1, this is the total buffer size summed across all
                        TP ranks. By default, this is set to None, which means
                        no KV offloading is enabled. When set, vLLM will
                        enable KV cache offloading to CPU using the
                        kv_offloading_backend. (default: None)
  --kv-sharing-fast-prefill, --no-kv-sharing-fast-prefill
                        This feature is work in progress and no prefill
                        optimization takes place with this flag enabled
                        currently.
                        In some KV sharing setups, e.g. YOCO
                        (https://arxiv.org/abs/2405.05254), some layers can
                        skip tokens corresponding to prefill. This flag
                        enables attention metadata for eligible layers to be
                        overridden with metadata necessary for implementing
                        this optimization in some models (e.g. Gemma3n)
                        (default: False)
  --mamba-block-size MAMBA_BLOCK_SIZE
                        Size of a contiguous cache block in number of tokens
                        for mamba cache. Can be set only when prefix caching
                        is enabled. Value must be a multiple of 8 to align
                        with causal_conv1d kernel. (default: None)
  --mamba-cache-dtype {auto,float16,float32}
                        The data type to use for the Mamba cache (both the
                        conv as well as the ssm state). If set to 'auto', the
                        data type will be inferred from the model config.
                        (default: auto)
  --mamba-cache-mode {align,all,none}
                        The cache strategy for Mamba layers. - "none": set
                        when prefix caching is disabled. - "all": cache the
                        mamba state of all tokens at position i * block_size.
                        This is  the default behavior (for models that support
                        it) when prefix caching is enabled. - "align": only
                        cache the mamba state of the last token of each
                        scheduler step and when the token is at position i *
                        block_size. (default: none)
  --mamba-ssm-cache-dtype {auto,float16,float32}
                        The data type to use for the Mamba cache (ssm state
                        only, conv state will still be controlled by
                        mamba_cache_dtype). If set to 'auto', the data type
                        for the ssm state will be determined by
                        mamba_cache_dtype. (default: auto)
  --num-gpu-blocks-override NUM_GPU_BLOCKS_OVERRIDE
                        Number of GPU blocks to use. This overrides the
                        profiled `num_gpu_blocks` if specified. Does nothing
                        if `None`. Used for testing preemption. (default:
                        None)
  --prefix-caching-hash-algo {sha256,sha256_cbor,xxhash,xxhash_cbor}
                        Set the hash algorithm for prefix caching:
                        - "sha256" uses Pickle for object serialization before
                        hashing. This is the current default, as SHA256 is the
                        most secure choice to avoid potential hash collisions.
                        - "sha256_cbor" provides a reproducible, cross-
                        language compatible hash. It serializes objects using
                        canonical CBOR and hashes them with SHA-256.
                        - "xxhash" uses Pickle serialization with xxHash
                        (128-bit) for faster, non-cryptographic hashing.
                        Requires the optional ``xxhash`` package. IMPORTANT:
                        Use of a hashing algorithm that is not considered
                        cryptographically secure theoretically increases the
                        risk of hash collisions, which can cause undefined
                        behavior or even leak private information in multi-
                        tenant environments. Even if collisions are still very
                        unlikely, it is important to consider your security
                        risk tolerance against the performance benefits before
                        turning this on.
                        - "xxhash_cbor" combines canonical CBOR serialization
                        with xxHash for reproducible hashing. Requires the
                        optional ``xxhash`` package. (default: sha256)

OffloadConfig:
  Configuration for model weight offloading to reduce GPU memory usage.

  --cpu-offload-gb CPU_OFFLOAD_GB
                        The space in GiB to offload to CPU, per GPU. Default
                        is 0, which means no offloading. Intuitively, this
                        argument can be seen as a virtual way to increase the
                        GPU memory size. For example, if you have one 24 GB
                        GPU and set this to 10, virtually you can think of it
                        as a 34 GB GPU. Then you can load a 13B model with
                        BF16 weight, which requires at least 26GB GPU memory.
                        Note that this requires fast CPU-GPU interconnect, as
                        part of the model is loaded from CPU memory to GPU
                        memory on the fly in each model forward pass. This
                        uses UVA (Unified Virtual Addressing) for zero-copy
                        access. (default: 0)
  --cpu-offload-params CPU_OFFLOAD_PARAMS [CPU_OFFLOAD_PARAMS ...]
                        The set of parameter name segments to target for CPU
                        offloading. Unmatched parameters are not offloaded. If
                        this set is empty, parameters are offloaded non-
                        selectively until the memory limit defined by
                        `cpu_offload_gb` is reached. Examples: - For parameter
                        name "mlp.experts.w2_weight": - "experts" or
                        "experts.w2_weight" will match. - "expert" or "w2"
                        will NOT match (must be exact segments). This allows
                        distinguishing parameters like "w2_weight" and
                        "w2_weight_scale". (default: set())
  --offload-backend {auto,prefetch,uva}
                        The backend for weight offloading. Options: - "auto":
                        Selects based on which sub-config has non-default
                        values (prefetch if offload_group_size > 0, uva if
                        cpu_offload_gb > 0). - "uva": UVA (Unified Virtual
                        Addressing) zero-copy offloading. - "prefetch": Async
                        prefetch with group-based layer offloading. (default:
                        auto)
  --offload-group-size OFFLOAD_GROUP_SIZE
                        Group every N layers together. Offload last
                        `offload_num_in_group` layers of each group. Default
                        is 0 (disabled). Example: group_size=8, num_in_group=2
                        offloads layers 6,7,14,15,22,23,... Unlike
                        cpu_offload_gb, this uses explicit async prefetching
                        to hide transfer latency. (default: 0)
  --offload-num-in-group OFFLOAD_NUM_IN_GROUP
                        Number of layers to offload per group. Must be <=
                        offload_group_size. Default is 1. (default: 1)
  --offload-params OFFLOAD_PARAMS [OFFLOAD_PARAMS ...]
                        The set of parameter name segments to target for
                        prefetch offloading. Unmatched parameters are not
                        offloaded. If this set is empty, ALL parameters of
                        each offloaded layer are offloaded. Uses segment
                        matching: "w13_weight" matches
                        "mlp.experts.w13_weight" but not
                        "mlp.experts.w13_weight_scale". (default: set())
  --offload-prefetch-step OFFLOAD_PREFETCH_STEP
                        Number of layers to prefetch ahead. Higher values hide
                        more latency but use more GPU memory. Default is 1.
                        (default: 1)

MultiModalConfig:
  Controls the behavior of multimodal models.

  --enable-mm-embeds, --no-enable-mm-embeds
                        If `True`, enables passing multimodal embeddings: for
                        `LLM` class, this refers to tensor inputs under
                        `multi_modal_data`; for the OpenAI-compatible server,
                        this refers to chat messages with content `"type":
                        "*_embeds"`.
                        When enabled with `--limit-mm-per-prompt` set to 0 for
                        a modality, precomputed embeddings skip count
                        validation for that modality,  saving memory by not
                        loading encoder modules while still enabling
                        embeddings as an input. Limits greater than 0 still
                        apply to embeddings.
                        WARNING: The vLLM engine may crash if incorrect shape
                        of embeddings is passed. Only enable this flag for
                        trusted users! (default: False)
  --interleave-mm-strings, --no-interleave-mm-strings
                        Enable fully interleaved support for multimodal
                        prompts, while using --chat-template-content-
                        format=string. (default: False)
  --language-model-only, --no-language-model-only
                        If True, disables all multimodal inputs by setting all
                        modality limits to 0. Equivalent to setting `--limit-
                        mm-per-prompt` to 0 for every modality. (default:
                        False)
  --limit-mm-per-prompt LIMIT_MM_PER_PROMPT
                        The maximum number of input items and options allowed
                        per prompt for each modality.
                        Defaults to 999 for each modality.
                        Legacy format (count only): {"image": 16, "video": 2}
                        Configurable format (with options): {"video":
                        {"count": 1, "num_frames": 32, "width": 512, "height":
                        512}, "image": {"count": 5, "width": 512, "height":
                        512}}
                        Mixed format (combining both): {"image": 16, "video":
                        {"count": 1, "num_frames": 32, "width": 512, "height":
                        512}}
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: {})
  --media-io-kwargs MEDIA_IO_KWARGS
                        Additional args passed to process media inputs, keyed
                        by modalities. For example, to set num_frames for
                        video, set `--media-io-kwargs '{"video":
                        {"num_frames": 40} }'`
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: {})
  --mm-encoder-attn-backend MM_ENCODER_ATTN_BACKEND
                        Optional override for the multi-modal encoder
                        attention backend when using vision transformers.
                        Accepts any value from `vllm.v1.attention.backends.reg
                        istry.AttentionBackendEnum` (e.g. `FLASH_ATTN`).
                        (default: None)
  --mm-encoder-only, --no-mm-encoder-only
                        When enabled, skips the language component of the
                        model.
                        This is usually only valid in disaggregated Encoder
                        process. (default: False)
  --mm-encoder-tp-mode {data,weights}
                        Indicates how to optimize multi-modal encoder
                        inference using tensor parallelism (TP).
                        - `"weights"`: Within the same vLLM engine, split the
                        weights of each layer across TP ranks. (default TP
                        behavior)
                        - `"data"`: Within the same vLLM engine, split the
                        batched input data across TP ranks to process the data
                        in parallel, while hosting the full weights on each TP
                        rank. This batch-level DP is not to be confused with
                        API request-level DP (which is controlled by `--data-
                        parallel-size`). This is only supported on a per-model
                        basis and falls back to `"weights"` if the encoder
                        does not support DP. (default: weights)
  --mm-processor-cache-gb MM_PROCESSOR_CACHE_GB
                        The size (in GiB) of the multi-modal processor cache,
                        which is used to avoid re-processing past multi-modal
                        inputs.
                        This cache is duplicated for each API process and
                        engine core process, resulting in a total memory usage
                        of `mm_processor_cache_gb * (api_server_count +
                        data_parallel_size)`.
                        Set to `0` to disable this cache completely (not
                        recommended). (default: 4)
  --mm-processor-cache-type {lru,shm}
                        Type of cache to use for the multi-modal
                        preprocessor/mapper. If `shm`, use shared memory FIFO
                        cache. If `lru`, use mirrored LRU cache. (default:
                        lru)
  --mm-processor-kwargs MM_PROCESSOR_KWARGS
                        Arguments to be forwarded to the model's processor for
                        multi-modal data, e.g., image processor. Overrides for
                        the multi-modal processor obtained from
                        `transformers.AutoProcessor.from_pretrained`.
                        The available overrides depend on the model that is
                        being run.
                        For example, for Phi-3-Vision: `{"num_crops": 4}`.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --mm-shm-cache-max-object-size-mb MM_SHM_CACHE_MAX_OBJECT_SIZE_MB
                        Size limit (in MiB) for each object stored in the
                        multi-modal processor shared memory cache. Only
                        effective when `mm_processor_cache_type` is `"shm"`.
                        (default: 128)
  --skip-mm-profiling, --no-skip-mm-profiling
                        When enabled, skips multimodal memory profiling and
                        only profiles with language backbone model during
                        engine initialization.
                        This reduces engine startup time but shifts the
                        responsibility to users for estimating the peak memory
                        usage of the activation of multimodal encoder and
                        embedding cache. (default: False)
  --video-pruning-rate VIDEO_PRUNING_RATE
                        Sets pruning rate for video pruning via Efficient
                        Video Sampling. Value sits in range [0;1) and
                        determines fraction of media tokens from each video to
                        be pruned. (default: None)

LoRAConfig:
  Configuration for LoRA.

  --default-mm-loras DEFAULT_MM_LORAS
                        Dictionary mapping specific modalities to LoRA model
                        paths; this field is only applicable to multimodal
                        models and should be leveraged when a model always
                        expects a LoRA to be active when a given modality is
                        present. Note that currently, if a request provides
                        multiple additional modalities, each of which have
                        their own LoRA, we do NOT apply default_mm_loras
                        because we currently only support one lora adapter per
                        prompt. When run in offline mode, the lora IDs for n
                        modalities will be automatically assigned to 1-n with
                        the names of the modalities in alphabetic order.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --enable-lora, --no-enable-lora
                        If True, enable handling of LoRA adapters. (default:
                        None)
  --enable-tower-connector-lora, --no-enable-tower-connector-lora
                        If `True`, LoRA support for the tower (vision encoder)
                        and connector  of multimodal models will be enabled.
                        This is an experimental feature and  currently only
                        supports some MM models such as the Qwen VL series.
                        The default  is False. (default: False)
  --fully-sharded-loras, --no-fully-sharded-loras
                        By default, only half of the LoRA computation is
                        sharded with tensor parallelism. Enabling this will
                        use the fully sharded layers. At high sequence length,
                        max rank or tensor parallel size, this is likely
                        faster. (default: False)
  --lora-dtype {auto,bfloat16,float16}
                        Data type for LoRA. If auto, will default to base
                        model dtype. (default: auto)
  --max-cpu-loras MAX_CPU_LORAS
                        Maximum number of LoRAs to store in CPU memory. Must
                        be >= than `max_loras`. (default: None)
  --max-lora-rank {1,8,16,32,64,128,256,320,512}
                        Max LoRA rank. (default: 16)
  --max-loras MAX_LORAS
                        Max number of LoRAs in a single batch. (default: 1)
  --specialize-active-lora, --no-specialize-active-lora
                        Whether to construct lora kernel grid by the number of
                        active LoRA adapters. When set to True, separate cuda
                        graphs will be captured for different counts of active
                        LoRAs (powers of 2 up to max_loras), which can improve
                        performance for variable LoRA usage patterns at the
                        cost of increased startup time and memory usage. Only
                        takes effect when cudagraph_specialize_lora is True.
                        (default: False)

ObservabilityConfig:
  Configuration for observability - metrics and tracing.

  --collect-detailed-traces {all,model,worker,None} [{all,model,worker,None} ...]
                        It makes sense to set this only if `--otlp-traces-
                        endpoint` is set. If set, it will collect detailed
                        traces for the specified modules. This involves use of
                        possibly costly and or blocking operations and hence
                        might have a performance impact.
                        Note that collecting detailed timing information for
                        each request can be expensive. (default: None)
  --cudagraph-metrics, --no-cudagraph-metrics
                        Enable CUDA graph metrics (number of padded/unpadded
                        tokens, runtime cudagraph dispatch modes, and their
                        observed frequencies at every logging interval).
                        (default: False)
  --enable-layerwise-nvtx-tracing, --no-enable-layerwise-nvtx-tracing
                        Enable layerwise NVTX tracing. This traces the
                        execution of each layer or module in the model and
                        attach information such as input/output shapes to nvtx
                        range markers. Noted that this doesn't work with CUDA
                        graphs enabled. (default: False)
  --enable-logging-iteration-details, --no-enable-logging-iteration-details
                        Enable detailed logging of iteration details. If set,
                        vllm EngineCore will log iteration details This
                        includes number of context/generation requests and
                        tokens and the elapsed cpu time for the iteration.
                        (default: False)
  --enable-mfu-metrics, --no-enable-mfu-metrics
                        Enable Model FLOPs Utilization (MFU) metrics.
                        (default: False)
  --kv-cache-metrics, --no-kv-cache-metrics
                        Enable KV cache residency metrics (lifetime, idle
                        time, reuse gaps). Uses sampling to minimize overhead.
                        Requires log stats to be enabled (i.e., --disable-log-
                        stats not set). (default: False)
  --kv-cache-metrics-sample KV_CACHE_METRICS_SAMPLE
                        Sampling rate for KV cache metrics (0.0, 1.0]. Default
                        0.01 = 1% of blocks. (default: 0.01)
  --otlp-traces-endpoint OTLP_TRACES_ENDPOINT
                        Target URL to which OpenTelemetry traces will be sent.
                        (default: None)
  --show-hidden-metrics-for-version SHOW_HIDDEN_METRICS_FOR_VERSION
                        Enable deprecated Prometheus metrics that have been
                        hidden since the specified version. For example, if a
                        previously deprecated metric has been hidden since the
                        v0.7.0 release, you use `--show-hidden-metrics-for-
                        version=0.7` as a temporary escape hatch while you
                        migrate to new metrics. The metric is likely to be
                        removed completely in an upcoming release. (default:
                        None)

SchedulerConfig:
  Scheduler configuration.

  --async-scheduling, --no-async-scheduling
                        If set to False, disable async scheduling. Async
                        scheduling helps to avoid gaps in GPU utilization,
                        leading to better latency and throughput. (default:
                        None)
  --disable-chunked-mm-input, --no-disable-chunked-mm-input
                        If set to true and chunked prefill is enabled, we do
                        not want to partially schedule a multimodal item. Only
                        used in V1 This ensures that if a request has a mixed
                        prompt (like text tokens TTTT followed by image tokens
                        IIIIIIIIII) where only some image tokens can be
                        scheduled (like TTTTIIIII, leaving IIIII), it will be
                        scheduled as TTTT in one step and IIIIIIIIII in the
                        next. (default: False)
  --disable-hybrid-kv-cache-manager, --no-disable-hybrid-kv-cache-manager
                        If set to True, KV cache manager will allocate the
                        same size of KV cache for all attention layers even if
                        there are multiple type of attention layers like full
                        attention and sliding window attention. If set to
                        None, the default value will be determined based on
                        the environment and starting configuration. (default:
                        None)
  --enable-chunked-prefill, --no-enable-chunked-prefill
                        If True, prefill requests can be chunked based on the
                        remaining `max_num_batched_tokens`.
                        The default value here is mainly for convenience when
                        testing. In real usage, this should be set in
                        `EngineArgs.create_engine_config`. (default: None)
  --long-prefill-token-threshold LONG_PREFILL_TOKEN_THRESHOLD
                        For chunked prefill, a request is considered long if
                        the prompt is longer than this number of tokens.
                        (default: 0)
  --max-long-partial-prefills MAX_LONG_PARTIAL_PREFILLS
                        For chunked prefill, the maximum number of prompts
                        longer than long_prefill_token_threshold that will be
                        prefilled concurrently. Setting this less than
                        max_num_partial_prefills will allow shorter prompts to
                        jump the queue in front of longer prompts in some
                        cases, improving latency. (default: 1)
  --max-num-batched-tokens MAX_NUM_BATCHED_TOKENS
                        Maximum number of tokens that can be processed in a
                        single iteration.
                        The default value here is mainly for convenience when
                        testing. In real usage, this should be set in
                        `EngineArgs.create_engine_config`.
                        Parse human-readable integers like '1k', '2M', etc.
                        Including decimal values with decimal multipliers.
                        Examples: - '1k' -> 1,000 - '1K' -> 1,024 - '25.6k' ->
                        25,600 (default: None)
  --max-num-partial-prefills MAX_NUM_PARTIAL_PREFILLS
                        For chunked prefill, the maximum number of sequences
                        that can be partially prefilled concurrently.
                        (default: 1)
  --max-num-seqs MAX_NUM_SEQS
                        Maximum number of sequences to be processed in a
                        single iteration.
                        The default value here is mainly for convenience when
                        testing. In real usage, this should be set in
                        `EngineArgs.create_engine_config`. (default: None)
  --scheduler-cls SCHEDULER_CLS
                        The scheduler class to use.
                        "vllm.v1.core.sched.scheduler.Scheduler" is the
                        default scheduler. Can be a class directly or the path
                        to a class of form "mod.custom_class". (default: None)
  --scheduling-policy {fcfs,priority}
                        The scheduling policy to use:
                        - "fcfs" means first come first served, i.e. requests
                        are handled in order of arrival.
                        - "priority" means requests are handled based on given
                        priority (lower value means earlier handling) and time
                        of arrival deciding any ties). (default: fcfs)
  --stream-interval STREAM_INTERVAL
                        The interval (or buffer size) for streaming in terms
                        of token length. A smaller value (1) makes streaming
                        smoother by sending each token immediately, while a
                        larger value (e.g., 10) reduces host overhead and may
                        increase throughput by batching multiple tokens before
                        sending. (default: 1)

CompilationConfig:
  Configuration for compilation.
  
      You must pass CompilationConfig to VLLMConfig constructor.
      VLLMConfig's post_init does further initialization. If used outside of the
      VLLMConfig, some fields will be left in an improper state.
  
      It contains PassConfig, which controls the custom fusion/transformation passes.
      The rest has three parts:
  
      - Top-level Compilation control:
          - [`mode`][vllm.config.CompilationConfig.mode]
          - [`debug_dump_path`][vllm.config.CompilationConfig.debug_dump_path]
          - [`cache_dir`][vllm.config.CompilationConfig.cache_dir]
          - [`backend`][vllm.config.CompilationConfig.backend]
          - [`custom_ops`][vllm.config.CompilationConfig.custom_ops]
          - [`splitting_ops`][vllm.config.CompilationConfig.splitting_ops]
          - [`compile_mm_encoder`][vllm.config.CompilationConfig.compile_mm_encoder]
      - CudaGraph capture:
          - [`cudagraph_mode`][vllm.config.CompilationConfig.cudagraph_mode]
          - [`cudagraph_capture_sizes`]
          [vllm.config.CompilationConfig.cudagraph_capture_sizes]
          - [`max_cudagraph_capture_size`]
          [vllm.config.CompilationConfig.max_cudagraph_capture_size]
          - [`cudagraph_num_of_warmups`]
          [vllm.config.CompilationConfig.cudagraph_num_of_warmups]
          - [`cudagraph_copy_inputs`]
          [vllm.config.CompilationConfig.cudagraph_copy_inputs]
      - Inductor compilation:
          - [`compile_sizes`][vllm.config.CompilationConfig.compile_sizes]
          - [`compile_ranges_endpoints`]
              [vllm.config.CompilationConfig.compile_ranges_endpoints]
          - [`inductor_compile_config`]
          [vllm.config.CompilationConfig.inductor_compile_config]
          - [`inductor_passes`][vllm.config.CompilationConfig.inductor_passes]
          - custom inductor passes
  
      Why we have different sizes for cudagraph and inductor:
      - cudagraph: a cudagraph captured for a specific size can only be used
          for the same size. We need to capture all the sizes we want to use.
      - inductor: a graph compiled by inductor for a general shape can be used
          for different sizes. Inductor can also compile for specific sizes,
          where it can have more information to optimize the graph with fully
          static shapes. However, we find the general shape compilation is
          sufficient for most cases. It might be beneficial to compile for
          certain small batchsizes, where inductor is good at optimizing.
      

  --cudagraph-capture-sizes CUDAGRAPH_CAPTURE_SIZES [CUDAGRAPH_CAPTURE_SIZES ...]
                        Sizes to capture cudagraph. - None (default): capture
                        sizes are inferred from vllm config. - list[int]:
                        capture sizes are specified as given. (default: None)
  --max-cudagraph-capture-size MAX_CUDAGRAPH_CAPTURE_SIZE
                        The maximum cudagraph capture size.
                        If cudagraph_capture_sizes is specified, this will be
                        set to the largest size in that list (or checked for
                        consistency if specified). If cudagraph_capture_sizes
                        is not specified, the list of sizes is generated
                        automatically following the pattern:
                        [1, 2, 4] + list(range(8, 256, 8)) + list( range(256,
                        max_cudagraph_capture_size + 1, 16))
                        If not specified, max_cudagraph_capture_size is set to
                        min(max_num_seqs*2, 512) by default. This voids OOM in
                        tight memory scenarios with small max_num_seqs, and
                        prevents capture of many large graphs (>512) that
                        would greatly increase startup time with limited
                        performance benefit. (default: None)

KernelConfig:
  Configuration for kernel selection and warmup behavior.

  --enable-flashinfer-autotune, --no-enable-flashinfer-autotune
                        If True, run FlashInfer autotuning during kernel
                        warmup. (default: None)
  --moe-backend {aiter,auto,cutlass,deep_gemm,flashinfer_cutedsl,flashinfer_cutlass,flashinfer_trtllm,marlin,triton}
                        Backend for MoE expert computation kernels. Available
                        options:
                        - "auto": Automatically select the best backend based
                        on model and hardware
                        - "triton": Use Triton-based fused MoE kernels
                        - "deep_gemm": Use DeepGEMM kernels (FP8 block-
                        quantized only)
                        - "cutlass": Use vLLM CUTLASS kernels
                        - "flashinfer_trtllm": Use FlashInfer with TRTLLM-GEN
                        kernels
                        - "flashinfer_cutlass": Use FlashInfer with CUTLASS
                        kernels
                        - "flashinfer_cutedsl": Use FlashInfer with CuteDSL
                        kernels (FP4 only)
                        - "marlin": Use Marlin kernels (weight-only
                        quantization)
                        - "aiter": Use AMD AITer kernels (ROCm only) (default:
                        auto)

VllmConfig:
  Dataclass which contains all vllm-related configuration. This
      simplifies passing around the distinct configurations in the codebase.
      

  --additional-config ADDITIONAL_CONFIG
                        Additional config for specified platform. Different
                        platforms may support different configs. Make sure the
                        configs are valid for the platform you are using.
                        Contents must be hashable. (default: {})
  --attention-config ATTENTION_CONFIG, -ac ATTENTION_CONFIG
                        Attention configuration.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default:
                        AttentionConfig(backend=None, flash_attn_version=None,
                        use_prefill_decode_attention=False,
                        flash_attn_max_num_splits_for_cuda_graph=32,
                        use_cudnn_prefill=False,
                        use_trtllm_ragged_deepseek_prefill=True,
                        use_trtllm_attention=None,
                        disable_flashinfer_prefill=False,
                        disable_flashinfer_q_quantization=False,
                        use_prefill_query_quantization=False))
  --compilation-config COMPILATION_CONFIG, -cc COMPILATION_CONFIG
                        `torch.compile` and cudagraph capture configuration
                        for the model.
                        As a shorthand, one can append compilation arguments
                        via -cc.parameter=argument such as `-cc.mode=3` (same
                        as `-cc='{"mode":3}'`).
                        You can specify the full compilation config like so:
                        `{"mode": 3, "cudagraph_capture_sizes": [1, 2, 4, 8]}`
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: {'mode': None,
                        'debug_dump_path': None, 'cache_dir': '',
                        'compile_cache_save_format': 'binary', 'backend':
                        'inductor', 'custom_ops': [], 'splitting_ops': None,
                        'compile_mm_encoder': False, 'compile_sizes': None,
                        'compile_ranges_endpoints': None,
                        'inductor_compile_config':
                        {'enable_auto_functionalized_v2': False,
                        'combo_kernels': True, 'benchmark_combo_kernel':
                        True}, 'inductor_passes': {}, 'cudagraph_mode': None,
                        'cudagraph_num_of_warmups': 0,
                        'cudagraph_capture_sizes': None,
                        'cudagraph_copy_inputs': False,
                        'cudagraph_specialize_lora': True,
                        'use_inductor_graph_partition': None, 'pass_config':
                        {}, 'max_cudagraph_capture_size': None,
                        'dynamic_shapes_config': {'type':
                        <DynamicShapesType.BACKED: 'backed'>,
                        'evaluate_guards': False, 'assume_32_bit_indexing':
                        False}, 'local_cache_dir': None,
                        'fast_moe_cold_start': None, 'static_all_moe_layers':
                        []})
  --ec-transfer-config EC_TRANSFER_CONFIG
                        The configurations for distributed EC cache transfer.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --kernel-config KERNEL_CONFIG
                        Kernel configuration.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default:
                        KernelConfig(enable_flashinfer_autotune=None,
                        moe_backend='auto'))
  --kv-events-config KV_EVENTS_CONFIG
                        The configurations for event publishing.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --kv-transfer-config KV_TRANSFER_CONFIG
                        The configurations for distributed KV cache transfer.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --optimization-level OPTIMIZATION_LEVEL
                        The optimization level. These levels trade startup
                        time cost for performance, with -O0 having the best
                        startup time and -O3 having the best performance. -O2
                        is used by default. See OptimizationLevel for full
                        description. (default: 2)
  --performance-mode {balanced,interactivity,throughput}
                        Performance mode for runtime behavior, 'balanced' is
                        the default. 'interactivity' favors low end-to-end
                        per-request latency at small batch sizes (fine-grained
                        CUDA graphs, latency-oriented kernels). 'throughput'
                        favors aggregate tokens/sec at high concurrency
                        (larger CUDA graphs, more aggressive batching,
                        throughput-oriented kernels). (default: balanced)
  --profiler-config PROFILER_CONFIG
                        Profiling configuration.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default:
                        ProfilerConfig(profiler=None, torch_profiler_dir='',
                        torch_profiler_with_stack=False,
                        torch_profiler_with_flops=False,
                        torch_profiler_use_gzip=True,
                        torch_profiler_dump_cuda_time_total=True,
                        torch_profiler_record_shapes=False,
                        torch_profiler_with_memory=False,
                        ignore_frontend=False, delay_iterations=0,
                        max_iterations=0, warmup_iterations=0,
                        active_iterations=5, wait_iterations=0))
  --speculative-config SPECULATIVE_CONFIG
                        Speculative decoding configuration.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)
  --structured-outputs-config STRUCTURED_OUTPUTS_CONFIG
                        Structured outputs configuration.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default:
                        StructuredOutputsConfig(backend='auto',
                        disable_any_whitespace=False,
                        disable_additional_properties=False,
                        reasoning_parser='', reasoning_parser_plugin='',
                        enable_in_reasoning=False))
  --weight-transfer-config WEIGHT_TRANSFER_CONFIG
                        The configurations for weight transfer during RL
                        training.
                        Should either be a valid JSON string or JSON keys
                        passed individually. (default: None)

When passing JSON CLI arguments, the following sets of arguments are equivalent:
   --json-arg '{"key1": "value1", "key2": {"key3": "value2"}}'
   --json-arg.key1 value1 --json-arg.key2.key3 value2

Additionally, list elements can be passed individually using +:
   --json-arg '{"key4": ["value3", "value4", "value5"]}'
   --json-arg.key4+ value3 --json-arg.key4+='value4,value5'
