diff --git a/litellm/router_strategy/lowest_cost.py b/litellm/router_strategy/lowest_cost.py
index b927df0c438..9e50122d69f 100644
--- a/litellm/router_strategy/lowest_cost.py
+++ b/litellm/router_strategy/lowest_cost.py
@@ -246,14 +246,21 @@ class LowestCostLoggingHandler(CustomLogger):
             )
             item_litellm_model_name = _deployment.get("litellm_params", {}).get("model")
             item_litellm_model_cost_map = litellm.model_cost.get(item_litellm_model_name, {})
+            if not item_litellm_model_cost_map and item_litellm_model_name:
+                try:
+                    item_litellm_model_cost_map = dict(
+                        litellm.get_model_info(model=item_litellm_model_name)
+                    )
+                except Exception:
+                    item_litellm_model_cost_map = {}
 
             # check if user provided input_cost_per_token and output_cost_per_token in litellm_params
             item_input_cost = None
             item_output_cost = None
-            if _deployment.get("litellm_params", {}).get("input_cost_per_token", None):
+            if _deployment.get("litellm_params", {}).get("input_cost_per_token") is not None:
                 item_input_cost = _deployment.get("litellm_params", {}).get("input_cost_per_token")
 
-            if _deployment.get("litellm_params", {}).get("output_cost_per_token", None):
+            if _deployment.get("litellm_params", {}).get("output_cost_per_token") is not None:
                 item_output_cost = _deployment.get("litellm_params", {}).get("output_cost_per_token")
 
             if item_input_cost is None:
diff --git a/tests/local_testing/test_lowest_cost_routing.py b/tests/local_testing/test_lowest_cost_routing.py
index 5bf3a3ee98b..9ec5d719cf1 100644
--- a/tests/local_testing/test_lowest_cost_routing.py
+++ b/tests/local_testing/test_lowest_cost_routing.py
@@ -46,6 +46,70 @@ async def test_get_available_deployments():
     assert selected_model["model_info"]["id"] == "groq-llama"
 
 
+@pytest.mark.asyncio
+async def test_get_available_deployments_provider_prefixed_model_pricing():
+    test_cache = DualCache()
+    model_list = [
+        {
+            "model_name": "gpt-3.5-turbo",
+            "litellm_params": {"model": "openai/gpt-4o-mini"},
+            "model_info": {"id": "prefixed-openai"},
+        },
+        {
+            "model_name": "gpt-3.5-turbo",
+            "litellm_params": {
+                "model": "azure/some-custom-deployment",
+                "input_cost_per_token": 0.5,
+                "output_cost_per_token": 0.5,
+            },
+            "model_info": {"id": "custom-priced"},
+        },
+    ]
+    lowest_cost_logger = LowestCostLoggingHandler(
+        router_cache=test_cache,
+    )
+
+    selected_model = await lowest_cost_logger.async_get_available_deployments(
+        model_group="gpt-3.5-turbo", healthy_deployments=model_list
+    )
+
+    assert selected_model["model_info"]["id"] == "prefixed-openai"
+
+
+@pytest.mark.asyncio
+async def test_get_available_deployments_zero_custom_price_is_honored():
+    test_cache = DualCache()
+    model_list = [
+        {
+            "model_name": "gpt-3.5-turbo",
+            "litellm_params": {
+                "model": "anthropic/claude-3-5-haiku-latest",
+                "input_cost_per_token": 0,
+                "output_cost_per_token": 0,
+            },
+            "model_info": {"id": "byok-zero-cost"},
+        },
+        {
+            "model_name": "gpt-3.5-turbo",
+            "litellm_params": {
+                "model": "azure/some-custom-deployment",
+                "input_cost_per_token": 0.5,
+                "output_cost_per_token": 0.5,
+            },
+            "model_info": {"id": "custom-priced"},
+        },
+    ]
+    lowest_cost_logger = LowestCostLoggingHandler(
+        router_cache=test_cache,
+    )
+
+    selected_model = await lowest_cost_logger.async_get_available_deployments(
+        model_group="gpt-3.5-turbo", healthy_deployments=model_list
+    )
+
+    assert selected_model["model_info"]["id"] == "byok-zero-cost"
+
+
 @pytest.mark.asyncio
 async def test_get_available_deployments_custom_price():
     from litellm._logging import verbose_router_logger