feat(api): support storing chat completions, enabling evals and model distillation in the dashboard (#1766)

Learn more at http://openai.com/devday2024

Author: stainless-app[bot]

.stats.yml

@@ -1,2 +1,2 @@
 configured_endpoints: 68
-openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/openai-8ad878332083dd506a478a293db78dc9e7b1b2124f2682e1d991225bc5bbcc3b.yml
+openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/openai-52b934aee6468039ec7f4ce046a282b5fbce114afc708e70f17121df654f71da.yml

src/openai/resources/chat/completions.py

@@ -30,7 +30,11 @@ class CompletionCreateParamsBase(TypedDict, total=False):
     messages: Required[Iterable[ChatCompletionMessageParam]]
     """A list of messages comprising the conversation so far.
 
-    [Example Python code](https://cookbook.openai.com/examples/how_to_format_inputs_to_chatgpt_models).
+    Depending on the [model](https://platform.openai.com/docs/models) you use,
+    different message types (modalities) are supported, like
+    [text](https://platform.openai.com/docs/guides/text-generation),
+    [images](https://platform.openai.com/docs/guides/vision), and
+    [audio](https://platform.openai.com/docs/guides/audio).
     """
 
     model: Required[Union[str, ChatModel]]
@@ -105,6 +109,12 @@ class CompletionCreateParamsBase(TypedDict, total=False):
     [o1 series models](https://platform.openai.com/docs/guides/reasoning).
     """
 
+    metadata: Optional[Dict[str, str]]
+    """
+    Developer-defined tags and values used for filtering completions in the
+    [dashboard](https://platform.openai.com/completions).
+    """
+
     n: Optional[int]
     """How many chat completion choices to generate for each input message.
 
@@ -183,6 +193,12 @@ class CompletionCreateParamsBase(TypedDict, total=False):
     stop: Union[Optional[str], List[str]]
     """Up to 4 sequences where the API will stop generating further tokens."""
 
+    store: Optional[bool]
+    """
+    Whether or not to store the output of this completion request for traffic
+    logging in the [dashboard](https://platform.openai.com/completions).
+    """
+
     stream_options: Optional[ChatCompletionStreamOptionsParam]
     """Options for streaming response. Only set this when you set `stream: true`."""
 

src/openai/types/chat/completion_create_params.py

@@ -12,6 +12,7 @@
     "gpt-4o",
     "gpt-4o-2024-08-06",
     "gpt-4o-2024-05-13",
+    "gpt-4o-realtime-preview-2024-10-01",
     "chatgpt-4o-latest",
     "gpt-4o-mini",
     "gpt-4o-mini-2024-07-18",

src/openai/types/chat_model.py

@@ -4,14 +4,25 @@
 
 from .._models import BaseModel
 
-__all__ = ["CompletionUsage", "CompletionTokensDetails"]
+__all__ = ["CompletionUsage", "CompletionTokensDetails", "PromptTokensDetails"]
 
 
 class CompletionTokensDetails(BaseModel):
+    audio_tokens: Optional[int] = None
+    """Audio input tokens generated by the model."""
+
     reasoning_tokens: Optional[int] = None
     """Tokens generated by the model for reasoning."""
 
 
+class PromptTokensDetails(BaseModel):
+    audio_tokens: Optional[int] = None
+    """Audio input tokens present in the prompt."""
+
+    cached_tokens: Optional[int] = None
+    """Cached tokens present in the prompt."""
+
+
 class CompletionUsage(BaseModel):
     completion_tokens: int
     """Number of tokens in the generated completion."""
@@ -24,3 +35,6 @@ class CompletionUsage(BaseModel):
 
     completion_tokens_details: Optional[CompletionTokensDetails] = None
     """Breakdown of tokens used in a completion."""
+
+    prompt_tokens_details: Optional[PromptTokensDetails] = None
+    """Breakdown of tokens used in the prompt."""

src/openai/types/completion_usage.py

@@ -56,13 +56,15 @@ def test_method_create_with_all_params_overload_1(self, client: OpenAI) -> None:
             logprobs=True,
             max_completion_tokens=0,
             max_tokens=0,
+            metadata={"foo": "string"},
             n=1,
             parallel_tool_calls=True,
             presence_penalty=-2,
             response_format={"type": "text"},
             seed=-9007199254740991,
             service_tier="auto",
             stop="string",
+            store=True,
             stream=False,
             stream_options={"include_usage": True},
             temperature=1,
@@ -177,13 +179,15 @@ def test_method_create_with_all_params_overload_2(self, client: OpenAI) -> None:
             logprobs=True,
             max_completion_tokens=0,
             max_tokens=0,
+            metadata={"foo": "string"},
             n=1,
             parallel_tool_calls=True,
             presence_penalty=-2,
             response_format={"type": "text"},
             seed=-9007199254740991,
             service_tier="auto",
             stop="string",
+            store=True,
             stream_options={"include_usage": True},
             temperature=1,
             tool_choice="none",
@@ -300,13 +304,15 @@ async def test_method_create_with_all_params_overload_1(self, async_client: Asyn
             logprobs=True,
             max_completion_tokens=0,
             max_tokens=0,
+            metadata={"foo": "string"},
             n=1,
             parallel_tool_calls=True,
             presence_penalty=-2,
             response_format={"type": "text"},
             seed=-9007199254740991,
             service_tier="auto",
             stop="string",
+            store=True,
             stream=False,
             stream_options={"include_usage": True},
             temperature=1,
@@ -421,13 +427,15 @@ async def test_method_create_with_all_params_overload_2(self, async_client: Asyn
             logprobs=True,
             max_completion_tokens=0,
             max_tokens=0,
+            metadata={"foo": "string"},
             n=1,
             parallel_tool_calls=True,
             presence_penalty=-2,
             response_format={"type": "text"},
             seed=-9007199254740991,
             service_tier="auto",
             stop="string",
+            store=True,
             stream_options={"include_usage": True},
             temperature=1,
             tool_choice="none",