From 25b2f83f97407d86c9616df99ee13b1c39f44157 Mon Sep 17 00:00:00 2001 From: Tin Chi Lo Date: Thu, 16 Jul 2026 19:26:28 -0700 Subject: [PATCH] test(router): clear the lru_cache when forcing the local cost map get_model_info is lru_cached, so swapping litellm.model_cost is not enough on its own. An earlier test that resolved these models against the remote map, which does not carry prompt_cache_min_tokens yet, leaves cached entries without it, and the stale hit resolves to the default. The assertions would then pass for the wrong reason or fail depending on execution order Clear on teardown as well, so entries these tests warm against the local map do not leak into later tests, matching the fixture already used in test_utils.py Also pin that a wildcard route resolves the underlying model's minimum. That works only because pattern_match_deployments substitutes the real model name into litellm_params before the deployment reaches the check; without the assertion that claim is unpinned and the threshold would silently fall back to the default --- .../test_prompt_caching_deployment_check.py | 34 ++++++++++++++++++- 1 file changed, 33 insertions(+), 1 deletion(-) diff --git a/tests/test_litellm/router_utils/pre_call_checks/test_prompt_caching_deployment_check.py b/tests/test_litellm/router_utils/pre_call_checks/test_prompt_caching_deployment_check.py index 1ad6caca2a3..6ad928b9737 100644 --- a/tests/test_litellm/router_utils/pre_call_checks/test_prompt_caching_deployment_check.py +++ b/tests/test_litellm/router_utils/pre_call_checks/test_prompt_caching_deployment_check.py @@ -26,9 +26,21 @@ def local_model_cost_map(monkeypatch): """ The remote cost map does not carry `prompt_cache_min_tokens` yet, so a test that reads the default map would pass here and flake in CI. Force the in-repo map. + + `get_model_info` is lru_cached, so swapping `model_cost` is not enough on its own: an earlier + test that resolved these models against the remote map leaves entries with no + `prompt_cache_min_tokens`, and the stale hit resolves to the default. Clear on the way out too, + so the entries these tests warm against the local map do not leak into later tests. """ + original_model_cost = litellm.model_cost monkeypatch.setenv("LITELLM_LOCAL_MODEL_COST_MAP", "True") - monkeypatch.setattr(litellm, "model_cost", litellm.get_model_cost_map(url="")) + litellm.model_cost = litellm.get_model_cost_map(url="") + litellm.get_model_info.cache_clear() + try: + yield + finally: + litellm.model_cost = original_model_cost + litellm.get_model_info.cache_clear() def _deployments(*models: str) -> List[dict]: @@ -156,3 +168,23 @@ async def test_async_filter_deployments_narrows_for_group_whose_model_minimum_is ) assert filtered == [deployments[1]] + + +@pytest.mark.asyncio +async def test_wildcard_route_resolves_underlying_model_minimum(local_model_cost_map): + from litellm import Router + + router = Router( + model_list=[ + { + "model_name": "anthropic/*", + "litellm_params": {"model": "anthropic/*", "api_key": "sk-fake"}, + "model_info": {"id": "wild-1"}, + } + ] + ) + + deployments = await router.async_get_healthy_deployments(model="anthropic/claude-opus-4-6", request_kwargs={}) + + assert deployments[0]["litellm_params"]["model"] == "anthropic/claude-opus-4-6" + assert _get_min_token_count_for_deployments(deployments) == 4096