How do I quickly get a binary file containing a binary response from the command line?

[Maferep]

If I do --download on the request it appears to generate an ascii encoded textfile, which is obviously not what I want, I want a real binary file. The response is an octet stream.

[amasen02]

When using the HTTPX CLI, the --download option requires a target file path:

httpx https://example.com/binary-endpoint --download output.bin

Under the hood, HTTPX opens this file in binary write mode (click.File("wb")) and streams raw bytes directly into it:

# httpx/_main.py:
for chunk in response.iter_bytes():
    download.write(chunk)

If you ran this and observed an ASCII text file instead of your expected binary stream, check for the two most common causes:

---

1. Cause: Server returned an HTTP error or HTML redirect

By default, --download writes whatever the server returns. If authentication failed, a token expired, or a redirect was not followed, the server returns an HTML/JSON error page (which is plain ASCII text):

• Add -v / --verbose or check the status code:

  httpx https://example.com/binary-endpoint -v --follow-redirects --download output.bin

• If the endpoint requires redirects, adding --follow-redirects ensures you download the final binary payload rather than an intermediate 302 Found redirect HTML page.

---

2. Cause: Piping stdout (--download - > file.bin)

If you passed - to download to standard output and piped it to a file:

# Don't do this with the CLI:
httpx https://example.com/binary-endpoint --download - > output.bin

The HTTPX CLI uses Rich to render an interactive progress bar (rich.progress.Progress). Terminal escape codes and progress bar characters can corrupt the binary stream when redirected through stdout.
Always supply an explicit file path:

httpx https://example.com/binary-endpoint --download output.bin

---

Alternative: Instant 1-line Python binary download

If you want guaranteed raw binary download with zero CLI progress-bar formatting or terminal interference:

python -c "import httpx, sys; r = httpx.get('https://example.com/binary-endpoint', follow_redirects=True); r.raise_for_status(); open('output.bin', 'wb').write(r.content)"

For large files, streaming avoids loading the whole payload into memory:

python -c "import httpx; client = httpx.Client(follow_redirects=True); 
with client.stream('GET', 'https://example.com/binary-endpoint') as r, open('output.bin', 'wb') as f:
    for chunk in r.iter_bytes(): f.write(chunk)"