openai/whisper · error · RuntimeError

{download_target} exists and is not a regular file

Error message

{download_target} exists and is not a regular file

What it means

Raised by whisper._download() before fetching a model checkpoint. The path where the library wants to store the model (download_root/<model>.pt) already exists but is not a regular file — typically a directory, a named pipe, or a dangling special file. The library refuses to overwrite non-file entries to avoid destroying data or writing into a directory.

Source

Thrown at whisper/__init__.py:61

    "medium.en": b"ABzY8usPae0{>%R7<zz_OvQ{)4kMa0BMw6u5rT}kRKX;$NfYBv00*Hl@qhsU00",
    "medium": b"ABzY8B0Jh+0{>%R7}kK1fFL7w6%<-Pf*t^=N)Qr&0RR9",
    "large-v1": b"ABzY8r9j$a0{>%R7#4sLmoOs{s)o3~84-RPdcFk!JR<kSfC2yj",
    "large-v2": b"ABzY8zd+h!0{>%R7=D0pU<_bnWW*tkYAhobTNnu$jnkEkXqp)j;w1Tzk)UH3X%SZd&fFZ2fC2yj",
    "large-v3": b"ABzY8gWO1E0{>%R7(9S+Kn!D~%ngiGaR?*L!iJG9p-nab0JQ=-{D1-g00",
    "large": b"ABzY8gWO1E0{>%R7(9S+Kn!D~%ngiGaR?*L!iJG9p-nab0JQ=-{D1-g00",
    "large-v3-turbo": b"ABzY8j^C+e0{>%RARaKHP%t(lGR*)0g!tONPyhe`",
    "turbo": b"ABzY8j^C+e0{>%RARaKHP%t(lGR*)0g!tONPyhe`",
}


def _download(url: str, root: str, in_memory: bool) -> Union[bytes, str]:
    os.makedirs(root, exist_ok=True)

    expected_sha256 = url.split("/")[-2]
    download_target = os.path.join(root, os.path.basename(url))

    if os.path.exists(download_target) and not os.path.isfile(download_target):
        raise RuntimeError(f"{download_target} exists and is not a regular file")

    if os.path.isfile(download_target):
        with open(download_target, "rb") as f:
            model_bytes = f.read()
        if hashlib.sha256(model_bytes).hexdigest() == expected_sha256:
            return model_bytes if in_memory else download_target
        else:
            warnings.warn(
                f"{download_target} exists, but the SHA256 checksum does not match; re-downloading the file"
            )

    with urllib.request.urlopen(url) as source, open(download_target, "wb") as output:
        with tqdm(
            total=int(source.info().get("Content-Length")),
            ncols=80,
            unit="iB",
            unit_scale=True,
            unit_divisor=1024,

View on GitHub (pinned to 5f86d1d863)

Solutions

1. Inspect the path: ls -la ~/.cache/whisper/ and check what <model>.pt actually is
2. If it is a directory containing the checkpoint, move the .pt file out of it or point load_model's download_root at the correct directory
3. Remove or rename the offending directory: rm -rf ~/.cache/whisper/<model>.pt (after confirming it holds nothing valuable), then retry load_model
4. Pass an explicit download_root to whisper.load_model pointing at an empty, writable directory

Example fix

# before
# ~/.cache/whisper/base.pt is a directory -> RuntimeError
model = whisper.load_model("base")

# after
import os, shutil
p = os.path.expanduser("~/.cache/whisper/base.pt")
if os.path.exists(p) and not os.path.isfile(p):
    shutil.rmtree(p)  # or move real checkpoint out first
model = whisper.load_model("base")

import os

def safe_download_root_ok(root: str, model: str) -> bool:
    target = os.path.join(root, f"{model}.pt")
    return not os.path.exists(target) or os.path.isfile(target)

try:
    model = whisper.load_model(name, download_root=root)
except RuntimeError as e:
    if "exists and is not a regular file" in str(e):
        # inspect/remove the offending path, then retry once with a clean root
        raise

When it happens

Trigger: Calling whisper.load_model('base') (or _download directly) when ~/.cache/whisper/base.pt (or $XDG_CACHE_HOME/whisper/base.pt, or a custom download_root) exists as a directory; e.g. someone previously ran 'mkdir -p ~/.cache/whisper/base.pt' or a bad symlink points to a directory.

Common situations: CI containers or shared servers where cache dirs were pre-created with wrong structure; users who unzipped a model into a folder named like the .pt file; symlinks from a shared model store pointing at a directory; a failed earlier tool that created a directory instead of a file.

Related errors

• Model {name} not found; available models = {available_models
• Model has been downloaded but the SHA256 checksum does not n
• Language {language} not found in tokenizer.
• Unsupported language: {language}
• model should be one of {available_models()} or path to a mod

AI-assisted analysis of openai/whisper@5f86d1d863 (2026-08-14). Data as JSON: /api/errors/2e1dfbdfe89d5a93. Report an issue: GitHub.