diff --git a/src/openai/resources/chat/completions/completions.py b/src/openai/resources/chat/completions/completions.py
index 5d56d05d87..eb13913e89 100644
--- a/src/openai/resources/chat/completions/completions.py
+++ b/src/openai/resources/chat/completions/completions.py
@@ -102,7 +102,7 @@ def parse(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         safety_identifier: str | Omit = omit,
         seed: Optional[int] | Omit = omit,
@@ -257,7 +257,7 @@ def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -564,7 +564,7 @@ def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -870,7 +870,7 @@ def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -1175,7 +1175,7 @@ def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -1454,7 +1454,7 @@ def stream(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         safety_identifier: str | Omit = omit,
         seed: Optional[int] | Omit = omit,
@@ -1598,7 +1598,7 @@ async def parse(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         safety_identifier: str | Omit = omit,
         seed: Optional[int] | Omit = omit,
@@ -1753,7 +1753,7 @@ async def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -2060,7 +2060,7 @@ async def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -2366,7 +2366,7 @@ async def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -2671,7 +2671,7 @@ async def create(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         response_format: completion_create_params.ResponseFormat | Omit = omit,
         safety_identifier: str | Omit = omit,
@@ -2950,7 +2950,7 @@ def stream(
         prediction: Optional[ChatCompletionPredictionContentParam] | Omit = omit,
         presence_penalty: Optional[float] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning_effort: Optional[ReasoningEffort] | Omit = omit,
         safety_identifier: str | Omit = omit,
         seed: Optional[int] | Omit = omit,
diff --git a/src/openai/resources/responses/responses.py b/src/openai/resources/responses/responses.py
index c85a94495d..b5b0fca368 100644
--- a/src/openai/resources/responses/responses.py
+++ b/src/openai/resources/responses/responses.py
@@ -127,7 +127,7 @@ def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -377,7 +377,7 @@ def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -626,7 +626,7 @@ def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -873,7 +873,7 @@ def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -976,7 +976,7 @@ def stream(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -1017,7 +1017,7 @@ def stream(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -1168,7 +1168,7 @@ def parse(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -1788,7 +1788,7 @@ async def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -2038,7 +2038,7 @@ async def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -2287,7 +2287,7 @@ async def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -2534,7 +2534,7 @@ async def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -2637,7 +2637,7 @@ def stream(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -2678,7 +2678,7 @@ def stream(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -2833,7 +2833,7 @@ async def parse(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -3918,7 +3918,7 @@ def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
@@ -3998,7 +3998,7 @@ async def create(
         previous_response_id: Optional[str] | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
         prompt_cache_key: str | Omit = omit,
-        prompt_cache_retention: Optional[Literal["in-memory", "24h"]] | Omit = omit,
+        prompt_cache_retention: Optional[Literal["in_memory", "24h"]] | Omit = omit,
         reasoning: Optional[Reasoning] | Omit = omit,
         safety_identifier: str | Omit = omit,
         service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]] | Omit = omit,
diff --git a/src/openai/types/chat/completion_create_params.py b/src/openai/types/chat/completion_create_params.py
index 8e71ccbe41..0379ee0865 100644
--- a/src/openai/types/chat/completion_create_params.py
+++ b/src/openai/types/chat/completion_create_params.py
@@ -185,7 +185,7 @@ class CompletionCreateParamsBase(TypedDict, total=False):
     [Learn more](https://platform.openai.com/docs/guides/prompt-caching).
     """
 
-    prompt_cache_retention: Optional[Literal["in-memory", "24h"]]
+    prompt_cache_retention: Optional[Literal["in_memory", "24h"]]
     """The retention policy for the prompt cache.
 
     Set to `24h` to enable extended prompt caching, which keeps cached prefixes
diff --git a/src/openai/types/responses/response.py b/src/openai/types/responses/response.py
index ada0783bce..0d2491ea7c 100644
--- a/src/openai/types/responses/response.py
+++ b/src/openai/types/responses/response.py
@@ -214,7 +214,7 @@ class Response(BaseModel):
     [Learn more](https://platform.openai.com/docs/guides/prompt-caching).
     """
 
-    prompt_cache_retention: Optional[Literal["in-memory", "24h"]] = None
+    prompt_cache_retention: Optional[Literal["in_memory", "24h"]] = None
     """The retention policy for the prompt cache.
 
     Set to `24h` to enable extended prompt caching, which keeps cached prefixes
diff --git a/src/openai/types/responses/response_create_params.py b/src/openai/types/responses/response_create_params.py
index bf7170da1f..a04495f40a 100644
--- a/src/openai/types/responses/response_create_params.py
+++ b/src/openai/types/responses/response_create_params.py
@@ -152,7 +152,7 @@ class ResponseCreateParamsBase(TypedDict, total=False):
     [Learn more](https://platform.openai.com/docs/guides/prompt-caching).
     """
 
-    prompt_cache_retention: Optional[Literal["in-memory", "24h"]]
+    prompt_cache_retention: Optional[Literal["in_memory", "24h"]]
     """The retention policy for the prompt cache.
 
     Set to `24h` to enable extended prompt caching, which keeps cached prefixes
diff --git a/src/openai/types/responses/responses_client_event.py b/src/openai/types/responses/responses_client_event.py
index 2bc6f899c5..5f9e73c61f 100644
--- a/src/openai/types/responses/responses_client_event.py
+++ b/src/openai/types/responses/responses_client_event.py
@@ -184,7 +184,7 @@ class ResponsesClientEvent(BaseModel):
     [Learn more](https://platform.openai.com/docs/guides/prompt-caching).
     """
 
-    prompt_cache_retention: Optional[Literal["in-memory", "24h"]] = None
+    prompt_cache_retention: Optional[Literal["in_memory", "24h"]] = None
     """The retention policy for the prompt cache.
 
     Set to `24h` to enable extended prompt caching, which keeps cached prefixes
diff --git a/src/openai/types/responses/responses_client_event_param.py b/src/openai/types/responses/responses_client_event_param.py
index 08596ef9ea..249c812116 100644
--- a/src/openai/types/responses/responses_client_event_param.py
+++ b/src/openai/types/responses/responses_client_event_param.py
@@ -185,7 +185,7 @@ class ResponsesClientEventParam(TypedDict, total=False):
     [Learn more](https://platform.openai.com/docs/guides/prompt-caching).
     """
 
-    prompt_cache_retention: Optional[Literal["in-memory", "24h"]]
+    prompt_cache_retention: Optional[Literal["in_memory", "24h"]]
     """The retention policy for the prompt cache.
 
     Set to `24h` to enable extended prompt caching, which keeps cached prefixes
diff --git a/tests/api_resources/chat/test_completions.py b/tests/api_resources/chat/test_completions.py
index 2b58ff8191..7b368d5a10 100644
--- a/tests/api_resources/chat/test_completions.py
+++ b/tests/api_resources/chat/test_completions.py
@@ -73,7 +73,7 @@ def test_method_create_with_all_params_overload_1(self, client: OpenAI) -> None:
             },
             presence_penalty=-2,
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning_effort="none",
             response_format={"type": "text"},
             safety_identifier="safety-identifier-1234",
@@ -207,7 +207,7 @@ def test_method_create_with_all_params_overload_2(self, client: OpenAI) -> None:
             },
             presence_penalty=-2,
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning_effort="none",
             response_format={"type": "text"},
             safety_identifier="safety-identifier-1234",
@@ -516,7 +516,7 @@ async def test_method_create_with_all_params_overload_1(self, async_client: Asyn
             },
             presence_penalty=-2,
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning_effort="none",
             response_format={"type": "text"},
             safety_identifier="safety-identifier-1234",
@@ -650,7 +650,7 @@ async def test_method_create_with_all_params_overload_2(self, async_client: Asyn
             },
             presence_penalty=-2,
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning_effort="none",
             response_format={"type": "text"},
             safety_identifier="safety-identifier-1234",
diff --git a/tests/api_resources/test_responses.py b/tests/api_resources/test_responses.py
index 8f82f2046d..d7e08f49f0 100644
--- a/tests/api_resources/test_responses.py
+++ b/tests/api_resources/test_responses.py
@@ -52,7 +52,7 @@ def test_method_create_with_all_params_overload_1(self, client: OpenAI) -> None:
                 "version": "version",
             },
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning={
                 "effort": "none",
                 "generate_summary": "auto",
@@ -139,7 +139,7 @@ def test_method_create_with_all_params_overload_2(self, client: OpenAI) -> None:
                 "version": "version",
             },
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning={
                 "effort": "none",
                 "generate_summary": "auto",
@@ -461,7 +461,7 @@ async def test_method_create_with_all_params_overload_1(self, async_client: Asyn
                 "version": "version",
             },
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning={
                 "effort": "none",
                 "generate_summary": "auto",
@@ -548,7 +548,7 @@ async def test_method_create_with_all_params_overload_2(self, async_client: Asyn
                 "version": "version",
             },
             prompt_cache_key="prompt-cache-key-1234",
-            prompt_cache_retention="in-memory",
+            prompt_cache_retention="in_memory",
             reasoning={
                 "effort": "none",
                 "generate_summary": "auto",