[Bug]: ModuleNotFoundError: No module named 'enterprise'

[whitfin]

What happened?

Importing the litellm.proxy (and probably other) modules crashes with the latest PyPI release. Worked up until the latest version of the package on PyPI.

Edit: introduced inside https://github.com/BerriAI/litellm/pull/10321.

Relevant log output

>>> import litellm.proxy
Traceback (most recent call last):
  File "<python-input-0>", line 1, in <module>
    import litellm.proxy
  File "/usr/local/lib/python3.13/site-packages/litellm/__init__.py", line 1058, in <module>
    from .responses.main import *
  File "/usr/local/lib/python3.13/site-packages/litellm/responses/main.py", line 13, in <module>
    from litellm.responses.litellm_completion_transformation.handler import (
        LiteLLMCompletionTransformationHandler,
    )
  File "/usr/local/lib/python3.13/site-packages/litellm/responses/litellm_completion_transformation/handler.py", line 8, in <module>
    from litellm.responses.litellm_completion_transformation.streaming_iterator import (
        LiteLLMCompletionStreamingIterator,
    )
  File "/usr/local/lib/python3.13/site-packages/litellm/responses/litellm_completion_transformation/streaming_iterator.py", line 5, in <module>
    from litellm.responses.litellm_completion_transformation.transformation import (
        LiteLLMCompletionResponsesConfig,
    )
  File "/usr/local/lib/python3.13/site-packages/litellm/responses/litellm_completion_transformation/transformation.py", line 9, in <module>
    from enterprise.enterprise_hooks.session_handler import (
    ...<2 lines>...
    )
ModuleNotFoundError: No module named 'enterprise'

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

v1.67.4

Twitter / LinkedIn details

No response

[DavdGao]

Similar problem for me.

[ishaan-jaff]

I've pushed a fix for this, it will be live on 1.67.4.post1

Please pin to 1.67.3 or lower in the meantime.

[MrOrz]

Note: 1.67.3 does not exist in pip index versions litellm:

pip3 index versions litellm
WARNING: pip index is currently an experimental command. It may be removed/changed in a future release without prior warning.
litellm (1.67.4)
Available versions: 1.67.4, 1.67.2, 1.67.1, 1.67.0.post1, 1.67.0, 1.66.3, 1.66.2, 1.66.1, 1.66.0, 1.65.8, 1.65.7, 1.65.6, 1.65.5, 1.65.4.post1 ...

So I pinned to 1.67.2 instead

[Naveed101633]

I also tried 1.67.3 but it's not working

[mmabrouk]

Same problem here. Solution was to pin to 1.67.2.

[SHREYASP21]

Is there a solution to this, downgrading is not helping

[mmabrouk]

Is there a solution to this, downgrading is not helping

Pinning to 1.67.2 fixes the issue. Make sure you are correctly doing it (maybe you are updating for instance pyproject.toml but not the lock file).

[SHREYASP21]

Hey, I’ve followed your advice and pinned litellm to version 1.67.2, but the issue still persists. I’m running this in Google Colab, where I simply install the package using pip install litellm==1.67.2 and try to call it immediately.

The ModuleNotFoundError for the enterprise module keeps occurring, and I don’t have any specific enterprise-related functionality in my script. Is there something else I’m missing? Any advice would be appreciated!

[shamalgithub]

Damn just when i wanted to use liteLLM it breaks 😂😂😂😂

litellm==1.67.4

has the same issue !

[Naveed101633]

Is there a solution to this, downgrading is not helping

Pinning to 1.67.2 fixes the issue. Make sure you are correctly doing it (maybe you are updating for instance pyproject.toml but not the lock file).

1.67.2 is not working, I checked it.

[SHREYASP21]

can someone explain what the issue is, I am new to using this tool, I basically pip installed litellm and called it, worked yesterday and isnt working now, what is the issue and can I not call it anymore

[MoshiBin]

Pinning to 1.67.2 works here. Used uv to install latest litellm, then told it to downgrade to 1.67.2:

$ uv tool install litellm[proxy]@1.67.2
Resolved 81 packages in 16.77s
      Built litellm-proxy-extras==0.1.11
Prepared 2 packages in 1.44s
Uninstalled 2 packages in 30ms
Installed 2 packages in 33ms
 - litellm==1.67.4
 + litellm==1.67.2
 - litellm-proxy-extras==0.1.12
 + litellm-proxy-extras==0.1.11
Installed 1 executable: litellm
$ litellm --config config.yaml
INFO:     Started server process [3157021]
INFO:     Waiting for application startup.

[ericmjl]

I can confirm that litellm==1.67.2 is working with no enterprise-related errors.

If you are still facing any enterprise errors, consider clearing out your virtual environment (venv, conda, pixi, uv, etc.), and reinstalling.

For those who are new, hang tight, litellm is awesome, these errors sometimes happen, and I'm sure the devs/maintainers (@krrishdholakia and @ishaan-jaff) know to add new integration tests in response.

[abooo96]

I can confirm too that downgrading to litellm==1.67.2 worked perfectly. Love your work guys!

[antonkulaga]

Lite LLM guys, PLEASE DO NOT BREAK STUFF in MINOR releases. I mean, if you want to move some essential modules to the enterprise, please first mark them somehow and do the official move in 1.68. I am among the numerous people who got everything broken just because of a minor litellm update.

[Livia-Zaharia]

I had similar issue with a series of collab notebooks I prepared for a workshop. Not sure how to fix to previous versions but can you give an estimate when a proper release that fixes this bug would be released? Fixing versions doesn't seem to be an option and most of our code uses liteLLM

[lorenzocesconetto]

I had the same problem with version 1.67.4. I can also confirm that pinning to a previous version solves the issue: litellm==1.67.2

[takeaship]

Downgrading to 1.67.2 also worked for uv environment. uv add litellm==1.67.2 --frozen

[ishaan-jaff]

Hi everyone, fix is live here: https://pypi.org/project/litellm/1.67.4.post1/

Apologies about the issue, I will be investigating further and posting an update with how we plan on preventing these types of issues and will link to testing improvements I push too.

[ishaan-jaff]

Lite LLM guys, PLEASE DO NOT BREAK STUFF in MINOR releases. I mean, if you want to move some essential modules to the enterprise, please first mark them somehow and do the official move in 1.68. I am among the numerous people who got everything broken just because of a minor litellm update.

hi @antonkulaga, acknowledging this feedback. I'm sorry about the issue and will do a better job around version upgrades/release conventions when we move files. Thank you for using litellm, we will do a better job on this.

[ericmjl]

@ishaan-jaff you guys have been awesome with LiteLLM. Keep up the good work!

[jamesbraza]

I've used https://github.com/Arkq/flake8-requirements to validate all imports are part of project requirements, maybe you can add that to the pre-commit toolchain here

[pseudotensor]

Same again happens with litellm 1.68.1. So much for not happening again.

[krrishdholakia]

We've just moved to using a separate pip package for enterprise (as of v1.69.0) - over the next week, we plan to move imports to use that (safer than expecting a folder to exist when it might not), and will close the issue once that work is completed.

[ishaan-jaff]

hi @pseudotensor do you have a stack trace of the exception ?

[ishaan-jaff]

@pseudotensor unable to repro the issue

• litellm version used 1.69.1 (latest)
• Here's a colab I tested with: https://colab.research.google.com/drive/1T5hzmK3UCFQ7SBw0JNYzieUgX44DKOoN?usp=sharing

[pseudotensor]

latest version works. So just another momentary blip.

(litellm) bash-5.2$ litellm --version
Traceback (most recent call last):
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/litellm/proxy/proxy_cli.py", line 507, in run_server
    from .proxy_server import (
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/litellm/proxy/proxy_server.py", line 224, in <module>
    from litellm.proxy.management_endpoints.internal_user_endpoints import (
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/litellm/proxy/management_endpoints/internal_user_endpoints.py", line 27, in <module>
    from litellm.proxy.hooks.user_management_event_hooks import UserManagementEventHooks
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/litellm/proxy/hooks/user_management_event_hooks.py", line 13, in <module>
    from enterprise.enterprise_callbacks.send_emails.base_email import BaseEmailLogger
ModuleNotFoundError: No module named 'enterprise'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/workspace/miniconda3/envs/litellm/bin/litellm", line 8, in <module>
    sys.exit(run_server())
             ^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/click/core.py", line 1161, in __call__
    return self.main(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/click/core.py", line 1082, in main
    rv = self.invoke(ctx)
         ^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/click/core.py", line 1443, in invoke
    return ctx.invoke(self.callback, **ctx.params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/click/core.py", line 788, in invoke
    return __callback(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/litellm/proxy/proxy_cli.py", line 519, in run_server
    from proxy_server import (
ModuleNotFoundError: No module named 'proxy_server'
(litellm) bash-5.2$ 

Can't even show the version due to the failure.

But:

(litellm) bash-5.2$ ls -alrt ~/miniconda3/envs/litellm/lib/python3.11/site-packages/|grep lite
drwxr-xr-x   2 root root    4096 May  9 20:03 litellm_proxy_extras-0.1.17.dist-info
drwxr-xr-x   4 root root    4096 May  9 20:03 litellm_proxy_extras
drwxr-xr-x  23 root root    4096 May  9 20:03 litellm
drwxr-xr-x   2 root root    4096 May  9 20:03 litellm-1.68.2.dist-info
(litellm) bash-5.2$ 

So also litellm 1.68.2 had issue.

[krrishdholakia]

Re-reading the error - this looks like it's been fixed since v1.67.4

File "/usr/local/lib/python3.13/site-packages/litellm/responses/litellm_completion_transformation/transformation.py", line 9, in from enterprise.enterprise_hooks.session_handler import ( ...<2 lines>... )

This line no longer exists in our codebase. It's now in a try-except.

Commit: https://github.com/BerriAI/litellm/commit/1e0ee2244543ff2fea0414467c71aca7813bde07#diff-ab79fe7e44da7e4b6266450180d908ea8d7dd6599bec809ef25d2bb5f31d4077

[krrishdholakia]

File "/workspace/miniconda3/envs/litellm/lib/python3.11/site-packages/litellm/proxy/hooks/user_management_event_hooks.py", line 13, in from enterprise.enterprise_callbacks.send_emails.base_email import BaseEmailLogger ModuleNotFoundError: No module named 'enterprise'

That's a separate import - which got moved in v1.69.0.

I believe there's probably some additional unaddressed cases for direct imports which we need to handle - will link the PR to this issue

[quinn-dougherty]

I've experimented with several of the versions mentioned in this thread, and I've noticed that attempting to use logfire callback is the culprit. litellm works fine if I don't try to integrate it with logfire.