openai/whisper · error · RuntimeError

Model {name} not found; available models = {available_models

Error message

Model {name} not found; available models = {available_models()}

What it means

whisper.load_model(name) only accepts (a) a key of the _MODELS registry (tiny, base, small, medium, large-v1/v2/v3, large-v3-turbo, turbo, and their .en variants) or (b) a path to an existing checkpoint file. Anything else raises RuntimeError listing the valid names.

Source

Thrown at whisper/__init__.py:143

    -------
    model : Whisper
        The Whisper ASR model instance
    """

    if device is None:
        device = "cuda" if torch.cuda.is_available() else "cpu"
    if download_root is None:
        default = os.path.join(os.path.expanduser("~"), ".cache")
        download_root = os.path.join(os.getenv("XDG_CACHE_HOME", default), "whisper")

    if name in _MODELS:
        checkpoint_file = _download(_MODELS[name], download_root, in_memory)
        alignment_heads = _ALIGNMENT_HEADS[name]
    elif os.path.isfile(name):
        checkpoint_file = open(name, "rb").read() if in_memory else name
        alignment_heads = None
    else:
        raise RuntimeError(
            f"Model {name} not found; available models = {available_models()}"
        )

    with (
        io.BytesIO(checkpoint_file) if in_memory else open(checkpoint_file, "rb")
    ) as fp:
        kwargs = {"weights_only": True} if torch.__version__ >= "1.13" else {}
        checkpoint = torch.load(fp, map_location=device, **kwargs)
    del checkpoint_file

    dims = ModelDimensions(**checkpoint["dims"])
    model = Whisper(dims)
    model.load_state_dict(checkpoint["model_state_dict"])

    if alignment_heads is not None:
        model.set_alignment_heads(alignment_heads)

    return model.to(device)

View on GitHub (pinned to 5f86d1d863)

Solutions

1. Print valid names: python -c "import whisper; print(whisper.available_models())" and use one of those exactly
2. For a local checkpoint, pass an absolute path: os.path.abspath(path), and confirm it exists first
3. Check spelling of the .en variants — the separator is a dot ('base.en', not 'base-en')
4. If the checkpoint is on another machine/symlink, verify os.path.isfile() resolves to True from the same process/CWD

Example fix

# before
model = whisper.load_model("medium-en")  # RuntimeError: Model not found

# after
import whisper, os
name = "medium.en"  # or a real path:
# name = os.path.abspath("checkpoints/finetuned.pt")
assert name in whisper.available_models() or os.path.isfile(name)
model = whisper.load_model(name)

import os, whisper

def is_loadable_model(name: str) -> bool:
    return name in whisper.available_models() or os.path.isfile(name)

from typing import Literal
ModelName = Literal["tiny", "base", "small", "medium", "large-v1", "large-v2", "large-v3", "large-v3-turbo", "turbo", "tiny.en", "base.en", "small.en", "medium.en"]

def is_model_name(v: str) -> "TypeGuard[ModelName]":
    import whisper
    return v in whisper.available_models()

try:
    model = whisper.load_model(name)
except RuntimeError as e:
    if "not found; available models" in str(e):
        raise ValueError(f"bad model {name!r}; valid: {whisper.available_models()}") from e
    raise

When it happens

Trigger: load_model('medium-en') (wrong separator, real name is 'medium.en'), load_model('whisper-large-v3'), or a relative path like load_model('models/base.pt') when the CWD is different — os.path.isfile(name) fails, so the name falls through to the error branch.

Common situations: Typos in model names (hyphen vs dot, missing .en), running a script from a different working directory with a relative checkpoint path, using a fine-tuned checkpoint path that does not exist yet, or assuming a model name from a different fork (e.g. faster-whisper/faster_distilrobust) exists here.

Related errors

• model should be one of {available_models()} or path to a mod
• {download_target} exists and is not a regular file
• Expected one of {set(str2val.keys())}, got {string}
• Language {language} not found in tokenizer.
• Unsupported language: {language}

AI-assisted analysis of openai/whisper@5f86d1d863 (2026-08-14). Data as JSON: /api/errors/d55663df38282c00. Report an issue: GitHub.