mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-11 03:38:38 +00:00
9fb55fe829
GET /v1/models was filtering by key.models and team.models but silently
ignoring LiteLLM_UserTable.models (the "Personal Models" field). A user
restricted to a subset of models would see the full proxy list on the
discovery endpoint even though calling any restricted model returned
401 at /v1/chat/completions — an inconsistency between listing and
inference enforcement.
Root cause: get_available_models_for_user() in litellm/proxy/utils.py
sources the model list from user_api_key_dict (key + team view) only.
The user object's models field is loaded by user_api_key_auth but
deliberately not spliced into UserAPIKeyAuth.
Fix: add get_user_models() + filter_models_by_user_access() helpers in
litellm/proxy/auth/model_checks.py mirroring the existing key/team
helpers, then call them as a final intersection step inside
get_available_models_for_user(). Defense-in-depth — the filter only
narrows what the key/team layer already permitted, never expands it.
Semantics match can_user_call_model() at inference time:
- user.models=[] -> no filter
- user.models=['no-default-models'] -> empty list
- user.models=['all-proxy-models'] -> no filter
- raw names / access groups -> exact + group expansion
- wildcards ('anthropic/*', '*') -> fnmatch.fnmatchcase
- user_id None (master/service key) -> no filter
- get_user_object failure -> no filter (swallow, log)
scope=expand admin branch in proxy_server.py untouched — intentional
admin escape hatch gated by _user_has_admin_privileges.
Tests (4 tiers, 28 new cases):
- Unit (test_model_checks.py): get_user_models + filter helper, 11 cases
- Async integration (test_proxy_utils.py): _apply_user_models_filter
across 6 branches incl. user_id None, no-default-models, all-proxy,
access groups, wildcard, exception swallow — 10 cases
- Route-level FastAPI TestClient
(discovery_endpoints/test_v1_models_user_filter.py): real model_list
handler → real get_available_models_for_user → mocked get_user_object,
7 cases
- E2E (e2e/cases/15): real proxy + Postgres, dynamic users via /user/new
+ /key/generate, verified PASS post-fix and FAIL pre-fix on the same
live proxy. Wired into e2e/tools/run-all-cases.
Closes internal mirror of BerriAI/litellm#26420.
|
||
|---|---|---|
| .. | ||
| agent_tests | ||
| audio_tests | ||
| basic_proxy_startup_tests | ||
| batches_tests | ||
| benchmarks | ||
| code_coverage_tests | ||
| documentation_tests | ||
| e2e | ||
| enterprise | ||
| guardrails_tests | ||
| image_gen_tests | ||
| integration | ||
| litellm | ||
| litellm-proxy-extras | ||
| litellm_core_utils | ||
| litellm_utils_tests | ||
| llm_responses_api_testing | ||
| llm_translation | ||
| load_tests | ||
| local_testing | ||
| logging_callback_tests | ||
| mcp_tests | ||
| multi_instance_e2e_tests | ||
| ocr_tests | ||
| old_proxy_tests/tests | ||
| openai_endpoints_tests | ||
| otel_tests | ||
| pass_through_tests | ||
| pass_through_unit_tests | ||
| proxy_admin_ui_tests | ||
| proxy_behavior | ||
| proxy_e2e_anthropic_messages_tests | ||
| proxy_migration_tests | ||
| proxy_security_tests | ||
| proxy_unit_tests | ||
| router_unit_tests | ||
| scim_tests | ||
| search_tests | ||
| spend_tracking_tests | ||
| store_model_in_db_tests | ||
| test_litellm | ||
| unified_google_tests | ||
| vector_store_tests | ||
| windows_tests | ||
| __init__.py | ||
| _fake_openai_endpoint_server.py | ||
| _flush_vcr_cache.py | ||
| _live_test_helpers.py | ||
| _openai_record_replay_proxy.py | ||
| _vcr_conftest_common.py | ||
| _vcr_redis_persister.py | ||
| eval_swe_bench.py | ||
| fake_openai_endpoint.py | ||
| gettysburg.wav | ||
| large_text.py | ||
| openai_batch_completions.jsonl | ||
| pyrightconfig.json | ||
| README.MD | ||
| test_anthropic_compaction_usage.py | ||
| test_budget_management.py | ||
| test_callbacks_on_proxy.py | ||
| test_config.py | ||
| test_debug_warning.py | ||
| test_default_encoding_non_root.py | ||
| test_end_users.py | ||
| test_entrypoint.py | ||
| test_fallbacks.py | ||
| test_gpt5_azure_temperature_support.py | ||
| test_health.py | ||
| test_keys.py | ||
| test_litellm_proxy_responses_config.py | ||
| test_logging.conf | ||
| test_models.py | ||
| test_new_vector_store_endpoints.py | ||
| test_openai_endpoints.py | ||
| test_organizations.py | ||
| test_otel_thread_leak.py | ||
| test_passthrough_endpoints.py | ||
| test_presidio_latency.py | ||
| test_proxy_server_non_root.py | ||
| test_ratelimit.py | ||
| test_resource_cleanup.py | ||
| test_service_logger_otel.py | ||
| test_spend_logs.py | ||
| test_team.py | ||
| test_team_logging.py | ||
| test_team_members.py | ||
| test_users.py | ||
In total litellm runs 1000+ tests
[02/20/2025] Update:
To make it easier to contribute and map what behavior is tested,
we've started mapping the litellm directory in tests/test_litellm
This folder can only run mock tests.