diff --git a/litellm/llms/openai/common_utils.py b/litellm/llms/openai/common_utils.py
index ffaf987d44e..505e1e45bdd 100644
--- a/litellm/llms/openai/common_utils.py
+++ b/litellm/llms/openai/common_utils.py
@@ -163,9 +163,10 @@ class OpenAISdkKeyParams(TypedDict, total=False):
 def openai_sdk_key_params(api_key: str | None) -> OpenAISdkKeyParams:
     if not _SDK_READS_ADMIN_KEY_ENV:
         return {"api_key": api_key}
-    if not api_key:
+    sdk_api_key: Final = os.environ.get("OPENAI_API_KEY") if api_key is None else api_key
+    if not sdk_api_key:
         raise openai.OpenAIError(_MISSING_OPENAI_API_KEY_MESSAGE)
-    return {"api_key": api_key, "admin_api_key": api_key}
+    return {"api_key": sdk_api_key, "admin_api_key": sdk_api_key}
 
 
 _DECODED_JSON: Final = TypeAdapter(object)
diff --git a/tests/integration/authorization/_responses_compact.py b/tests/integration/authorization/_responses_compact.py
index 8270d729667..e33aa749c42 100644
--- a/tests/integration/authorization/_responses_compact.py
+++ b/tests/integration/authorization/_responses_compact.py
@@ -322,7 +322,7 @@ def assert_compacted(compacted: CompactedResponse, model: str, text: str) -> Non
                 "encrypted_content": COMPACT_STATE,
             }
         ],
-        "usage": dict(COMPACT_USAGE),
+        "usage": {**COMPACT_USAGE, "input_tokens_details": {"cache_write_tokens": 0, "cached_tokens": 0}},
     }, compacted.model_dump_json()
 
 
diff --git a/tests/integration/providers/test_responses_compact_wire.py b/tests/integration/providers/test_responses_compact_wire.py
index 88fbec27c07..0c446d8dcc5 100644
--- a/tests/integration/providers/test_responses_compact_wire.py
+++ b/tests/integration/providers/test_responses_compact_wire.py
@@ -282,7 +282,7 @@ def test_sdk_compact_reaches_the_compact_endpoint_with_its_full_body(gateway: Ga
             "output": [output_item],
             "usage": {
                 "input_tokens": 7,
-                "input_tokens_details": {"cached_tokens": 0},
+                "input_tokens_details": {"cache_write_tokens": 0, "cached_tokens": 0},
                 "output_tokens": 3,
                 "output_tokens_details": {"reasoning_tokens": 0},
                 "total_tokens": 10,
diff --git a/tests/integration/translation/responses/bases/gemini.py b/tests/integration/translation/responses/bases/gemini.py
index 597dbc25ab7..16915b16079 100644
--- a/tests/integration/translation/responses/bases/gemini.py
+++ b/tests/integration/translation/responses/bases/gemini.py
@@ -577,6 +577,7 @@ GEMINI_3_5_FLASH_TEST_CASE: Final = TranslationTestCase(
                 "image_tokens": None,
                 "text_tokens": 10,
                 "video_tokens": None,
+                "cache_write_tokens": 0,
             },
             "output_tokens": 127,
             "output_tokens_details": {"audio_tokens": None, "reasoning_tokens": 125, "text_tokens": 2},
diff --git a/tests/integration/translation/responses/bases/vertex_ai.py b/tests/integration/translation/responses/bases/vertex_ai.py
index 59637e380ce..b68eec3f06b 100644
--- a/tests/integration/translation/responses/bases/vertex_ai.py
+++ b/tests/integration/translation/responses/bases/vertex_ai.py
@@ -472,6 +472,7 @@ GEMINI_3_5_FLASH_TEST_CASE: Final = TranslationTestCase(
                 "image_tokens": None,
                 "text_tokens": 9,
                 "video_tokens": None,
+                "cache_write_tokens": 0,
             },
             "output_tokens": 86,
             "output_tokens_details": {"audio_tokens": None, "reasoning_tokens": 84, "text_tokens": 2},
diff --git a/tests/unit/llms/azure/test_azure_common_utils.py b/tests/unit/llms/azure/test_azure_common_utils.py
index ecd2823b503..260f61ab838 100644
--- a/tests/unit/llms/azure/test_azure_common_utils.py
+++ b/tests/unit/llms/azure/test_azure_common_utils.py
@@ -11,6 +11,7 @@ from unittest.mock import ANY, MagicMock, patch
 
 import httpx
 import pytest
+import respx
 from openai import AsyncAzureOpenAI
 
 import litellm
@@ -2652,3 +2653,37 @@ def test_azure_region_lookup_sends_the_key_and_bearer_headers(monkeypatch: pytes
     params: Final = LiteLLM_Params(model="azure/gpt-region", api_key="azure-key", api_base=api_base)
 
     assert get_model_region(litellm_params=params, mode="chat") == "eastus"
+
+
+@pytest.mark.parametrize("is_async", [False, True])
+@pytest.mark.asyncio
+async def test_azure_v1_deployment_keyed_only_by_the_openai_api_key_env_sends_that_key(
+    monkeypatch: pytest.MonkeyPatch, respx_mock: respx.MockRouter, is_async: bool
+) -> None:
+    for name in ("AZURE_API_KEY", "AZURE_OPENAI_API_KEY", "AZURE_AD_TOKEN", "AZURE_OPENAI_AD_TOKEN", "AZURE_CLIENT_ID"):
+        monkeypatch.delenv(name, raising=False)
+    monkeypatch.setenv("OPENAI_API_KEY", "azure-key-in-openai-env")
+    monkeypatch.setenv("OPENAI_ADMIN_KEY", "sk-admin-key-for-org-apis")
+    monkeypatch.setenv("DISABLE_AIOHTTP_TRANSPORT", "True")
+    api_base: Final = f"https://envkey-{is_async}.openai.azure.com"
+    route: Final = respx_mock.post(f"{api_base}/openai/v1/chat/completions").respond(
+        200,
+        json={
+            "id": "chatcmpl-envkey",
+            "object": "chat.completion",
+            "created": 0,
+            "model": "gpt-4o-mini",
+            "choices": [{"index": 0, "finish_reason": "stop", "message": {"role": "assistant", "content": "hi"}}],
+        },
+    )
+    request: Final = {
+        "model": "azure/gpt-4o-mini",
+        "messages": [{"role": "user", "content": "hi"}],
+        "api_base": api_base,
+        "api_version": "v1",
+    }
+
+    response: Final = await litellm.acompletion(**request) if is_async else litellm.completion(**request)
+
+    assert response.choices[0].message.content == "hi"
+    assert route.calls.last.request.headers["authorization"] == "Bearer azure-key-in-openai-env"