fix(proxy): restore wildcard expansion in /v1/model/info

[Sameerlite]

Summary

• Adds expand_wildcard_deployments_for_model_info() to model_checks.py that fans out wildcard deployments (e.g. openai/*, *) into one row per known provider model, restoring the behaviour users expected.
• The expansion is applied only to the /v1/model/info endpoint (model_info_v1). /v2/model/info is unchanged.

---

Note

Medium Risk
Touches model listing and permission filtering for a widely used admin/API endpoint; logic is defensive but expansion can increase response size for broad wildcards.

Overview
Restores wildcard fan-out on /v1/model/info after deployments started coming from llm_router.model_list (which fixed team-scoped rows but dropped per-model expansion).

Adds expand_wildcard_deployments_for_model_info(), which duplicates wildcard deployment dicts into one row per known provider model (via get_known_models_from_wildcard), aligned with /v1/models. /v2/model/info is unchanged.

Expansion picks the wildcard from model_name vs litellm_params.model so a concrete alias is not replaced when only the backend model is wildcarded; invalid LiteLLM_Params or empty expansions pass through unchanged to avoid errors.

Unit tests cover passthrough, openai/* expansion, alias preservation, and bad params.

^{Reviewed by Cursor Bugbot for commit c00185b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[greptile-apps]

Greptile Summary

Restores wildcard fan-out on /v1/model/info by introducing expand_wildcard_deployments_for_model_info() in model_checks.py, called once per request before the permission filter. /v2/model/info is unchanged.

• Wildcard deployments (e.g. model_name=\"*\" / litellm_params.model=\"openai/*\") are fanned out into one row per known provider model, matching the behaviour of /v1/models; the branch ordering ensures a concrete alias model_name is never overwritten.
• LiteLLM_Params construction is wrapped in try/except, and empty-expansion cases pass through unchanged, making the function fully defensive against unusual stored params.
• Four unit tests and one integration test cover the main scenarios (passthrough, openai/* expansion, alias preservation, invalid params).

Confidence Score: 5/5

The change is additive and fully defensive — empty expansions and bad params both fall through unchanged — with no modifications to existing auth or routing logic.

The core expansion logic is correctly ordered and guarded; the only open items are a minor documentation gap in one test and a note about downstream dict-vs-object reads for expanded rows, neither of which affects runtime behaviour today.

No files require special attention; the integration test in test_routes_model_info.py only exercises unrestricted users, so restricted-user coverage would be a worthwhile follow-up.

Important Files Changed

Filename	Overview
litellm/proxy/auth/model_checks.py	Adds expand_wildcard_deployments_for_model_info with well-ordered branching logic, defensive try/except around LiteLLM_Params construction, and correct deepcopy of expanded rows.
litellm/proxy/proxy_server.py	Single-line insertion of expand_wildcard_deployments_for_model_info call correctly placed before allowed_model_names filtering, so expansion is visible to the permission check.
tests/test_litellm/proxy/auth/test_model_checks.py	Four new unit tests cover passthrough, openai/* expansion, alias preservation, and invalid-params passthrough; the invalid-params test has an implicit dependency on internal Pydantic coercion that should be documented.
tests/test_litellm/proxy/proxy_server/test_routes_model_info.py	Integration test validates star-wildcard expansion survives the deployment filter for unrestricted users; restricted-user (team/key scoped) interaction is not tested.

_{Reviews (5): Last reviewed commit: "fix(proxy): fix reportUnknownArgumentTyp..." | Re-trigger Greptile}

[codecov]

Codecov Report

❌ Patch coverage is 91.89189% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
litellm/proxy/auth/model_checks.py	91.66%	3 Missing ⚠️

📢 Thoughts on this report? Let us know!

[Sameerlite]

@greptileai

[Sameerlite]

@greptileai

[Sameerlite]

bugbot run

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

Bugbot Autofix prepared a fix for the issue found in the latest run.

• ✅ Fixed: Star wildcard rows dropped after filter
  • Expanded allowlists now use the provider wildcard from litellm_params when the public route is '*', so scoped model-info filters retain the concrete rows.

_{You can send follow-ups to the cloud agent here.}

^{Reviewed by Cursor Bugbot for commit c00185b. Configure here.}

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ Sameerlite
❌ cursoragent
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[Sameerlite]

@greptileai

[Sameerlite]

@greptileai

[mateo-berri]

LGTM; thanks!