ERROR: Could not find a version that satisfies the requirement — how to fix

ERROR: Could not find a version that satisfies the requirement tensorflow (from versions: none)
ERROR: No matching distribution found for tensorflow

Quick fix: Check the exact name on pypi.org, check your Python version (python -V) against the package's supported versions, upgrade pip (python -m pip install -U pip), and run python -m pip index versions <pkg> to see what's available. Use a supported Python (e.g. 3.12 instead of a brand-new release) in a venv if needed.

What it means

pip downloaded the list of files for the package from the index (PyPI by default) and none of them matched your environment and requirement. The (from versions: ...) list shows which versions pip considered compatible. If it says none, nothing at all was usable.

First, gather facts:

python -V
python -m pip -V                       # which Python is pip tied to?
python -m pip index versions requests  # what versions are visible (pip 21.2+)

Cause 1: wrong package name

The import name is often not the PyPI name:

Standard-library modules (json, os, sqlite3, tkinter) can't be installed with pip at all.

Cause 2: your Python version isn't supported

Packages with compiled extensions (tensorflow, torch, numpy older versions, etc.) publish wheels only for specific Python versions. On a brand-new Python release, or on an old one, there may be no matching files. Check "Requires: Python" and the file list on the package's pypi.org page.

Fix: create a venv with a supported interpreter:

python3.12 -m venv .venv
source .venv/bin/activate
python -m pip install -U pip
pip install tensorflow

Also check you're not on 32-bit Python on Windows — many packages ship 64-bit wheels only.

Cause 3: the pinned version doesn't exist or is too old

ERROR: Could not find a version that satisfies the requirement django==5.9 (from versions: 1.1.3, ..., 5.1.2)

Here the list is populated — pick a version from it, or loosen the pin in requirements.txt (e.g. django>=5.0,<6). Some old releases also declare Requires-Python that excludes your interpreter, so they're hidden from the list.

Cause 4: outdated pip

Old pip versions don't understand newer wheel tags (e.g. manylinux_2_28, macOS arm64) and skip valid files:

python -m pip install --upgrade pip

Cause 5: no access to the index (network, proxy, SSL)

If every package fails with (from versions: none), pip probably can't reach PyPI. Look for warnings above the error like Retrying ... NewConnectionError or SSLError.

python -m pip install requests -v          # verbose: shows URLs tried
python -m pip config list                  # custom index-url set?
pip install --proxy http://proxy:8080 requests

A misconfigured index-url in pip.conf / PIP_INDEX_URL pointing at a private mirror that lacks the package gives the same error. Add PyPI explicitly with --extra-index-url https://pypi.org/simple if your policy allows it.

Cause 6: package needs a special index

PyTorch CUDA builds are on PyTorch's own index; follow the command from pytorch.org, e.g.:

pip install torch --index-url https://download.pytorch.org/whl/cu121

Cause 7: pre-release only

If a package only has pre-releases (2.0.0b1), pip ignores them unless you ask: pip install --pre <pkg>.

Last updated 2026-10-30.