How to check the Haystack version

Haystack projects often run inside virtual environments, notebook kernels, or service containers where a global Python package is not the package the application imports. Checking the version from the same interpreter used by the project shows which haystack-ai distribution is active.

The current Haystack 2.x package is distributed as haystack-ai and imported in Python as haystack. Reading the distribution metadata through Python avoids mismatches between a shell-level pip command and the interpreter that runs the project.

Use python -m pip for follow-up package metadata so pip stays tied to that same interpreter. Older Haystack 1.x environments used farm-haystack, so a migration check should identify that package separately instead of treating it as a haystack-ai version.

1. Open the Python environment that runs the Haystack project.

   Activate the project virtual environment, select the project notebook kernel, or open the service container before checking the package.
   Related: How to create a virtualenv for Haystack

2. Print the active interpreter and the installed haystack-ai distribution version.

   $ python -c 'import sys, importlib.metadata as md; print("python=" + sys.executable); print("haystack-ai=" + md.version("haystack-ai"))'
   python=/home/user/haystack-demo/.venv/bin/python
   haystack-ai=2.30.2

   If this raises PackageNotFoundError, the active interpreter cannot see the haystack-ai distribution. Install it in this environment or switch to the interpreter that owns the package.
   Related: How to install Haystack with pip

3. Confirm where the haystack import loads from.

   $ python -c 'import haystack; print("haystack=" + haystack.__file__); print("haystack.__version__=" + haystack.__version__)'
   haystack=/home/user/haystack-demo/.venv/lib/python3.14/site-packages/haystack/__init__.py
   haystack.__version__=2.30.2

   The import path should point inside the same environment as the interpreter path above, such as the project virtual environment or container path.

4. Read package metadata through the same interpreter.

   $ python -m pip show haystack-ai
   Name: haystack-ai
   Version: 2.30.2
   ##### snipped #####
   Location: /home/user/haystack-demo/.venv/lib/python3.14/site-packages
   Requires: docstring-parser, filetype, haystack-experimental, httpx, jinja2
   ##### snipped #####
   Required-by: haystack-experimental

   python -m pip keeps the metadata check bound to the interpreter that printed the version, which avoids shell pip aliases from another environment.

5. Check for the legacy farm-haystack package when imports or migration history look suspicious.

   $ python -m pip show farm-haystack
   WARNING: Package(s) not found: farm-haystack

   farm-haystack and haystack-ai are different distributions. If both appear in one environment, clean up the migrated environment before trusting version or import checks.
   Related: How to migrate Haystack 1.x code to Haystack 2.x