vllm-project/vllm · error · ValueError

Expected model after 'vllm serve', got {model!r}

Error message

Expected model after 'vllm serve', got {model!r}

What it means

After validating the ['vllm','serve'] prefix, argv_to_config() checks that argv[2] (the model) does not start with '-'. A leading dash means the token is a flag, so the positional model is missing — e.g. ['vllm','serve','--port','8000'] with the model forgotten or placed later.

Source

Thrown at tools/recipes/recipe_json_to_vllm_config.py:470

    Nested JSON field names after a dot are preserved.
    """
    parts = raw_key.split(".")
    parts[0] = parts[0].replace("_", "-")
    return parts


def argv_to_config(argv: list[Any]) -> dict[str, Any]:
    argv = [str(x) for x in argv]

    if len(argv) < 3 or argv[0:2] != ["vllm", "serve"]:
        raise ValueError(
            "Expected recipe argv to start with: ['vllm', 'serve', MODEL, ...]. "
            f"Got: {argv[:4]!r}"
        )

    model = argv[2]
    if model.startswith("-"):
        raise ValueError(f"Expected model after 'vllm serve', got {model!r}")

    config: dict[str, Any] = {"model": model}

    i = 3
    while i < len(argv):
        token = argv[i]

        # -O3 / -O=3
        if token.startswith("-O") and token != "-O":
            value = token[3:] if token.startswith("-O=") else token[2:]
            merge_value(config, ["optimization-level"], coerce(value))
            i += 1
            continue

        # -O 3
        if token == "-O":
            if i + 1 >= len(argv):
                raise ValueError("-O is missing its value")

View on GitHub (pinned to c794754062)

Solutions

1. jq '.argv' recipe.json and look at argv[2]
2. Insert the model id at position 2 (right after 'serve') in the recipe JSON
3. If the recipe is API-rendered with the model missing, report the model's recipe as broken and pick another model/hardware recipe

Example fix

# before
"argv": ["vllm", "serve", "--max-model-len", "8192"]
# after
"argv": ["vllm", "serve", "meta-llama/Llama-3-8B", "--max-model-len", "8192"]

argv = [str(x) for x in recipe["argv"]]
if len(argv) > 2 and argv[2].startswith("-"):
    raise SystemExit("argv[2] must be the model id, got a flag")

def argv_has_model(argv: list) -> bool:
    argv = [str(x) for x in argv]
    return len(argv) >= 3 and not argv[2].startswith("-")

When it happens

Trigger: Recipe rendering that emits flags before the model; model field empty and dropped from argv leaving a flag at index 2; hand-built argv where the model was omitted by accident.

Common situations: Recipes renderer bug for models with empty ids; editing recipe JSON and deleting the model token; schema where the model is supposed to be supplied separately but argv still contains only options.

Related errors

• Expected recipe argv to start with: ['vllm', 'serve', MODEL,
• Unexpected positional/short argument {token!r}. The converte
• Cannot merge nested option {'.'.join(path)!r}: {part!r} is a
• -O is missing its value
• {token} is missing its value

AI-assisted analysis of vllm-project/vllm@c794754062 (2026-08-14). Data as JSON: /api/errors/39692cf9121965ee. Report an issue: GitHub.