feat: update Chat Completions request/response schemas

README.md

@@ -25,6 +25,7 @@
   - [Make Chat Completions Requests](#make-chat-completions-requests)
     - [Without Streaming](#without-streaming)
     - [With Streaming](#with-streaming)
+    - [DIAL-specific and Extended Parameters](#dial-specific-and-extended-parameters)
   - [Working with Files](#working-with-files)
     - [Working with URLs](#working-with-urls)
     - [Uploading Files](#uploading-files)
@@ -464,6 +465,115 @@ ChatCompletionChunk(
 )
 ```
 
+#### DIAL-specific and Extended Parameters
+
+Along with the standard OpenAI parameters, `chat.completions.create` accepts
+the DIAL extensions and the newer OpenAI parameters:
+
+```python
+completion = client.chat.completions.create(
+    deployment_name="gpt-4o",
+    stream=False,
+    messages=[
+        # Messages support multi-modal content parts,
+        # the "developer" role and per-message cache breakpoints
+        {
+            "role": "developer",
+            "content": "Be brief",
+            "custom_fields": {"cache_breakpoint": {"expire_at": "1h"}},
+        },
+        {
+            "role": "user",
+            "content": [
+                {"type": "text", "text": "What is on the picture?"},
+                {
+                    "type": "image_url",
+                    "image_url": {"url": "https://example.com/image.png"},
+                },
+            ],
+            # DIAL attachments, stages and forms
+            "custom_content": {
+                "attachments": [
+                    {"type": "image/png", "url": "files/bucket/image.png"}
+                ]
+            },
+        },
+    ],
+    tools=[
+        {
+            "type": "function",
+            "function": {"name": "get_weather", "parameters": {}, "strict": True},
+            "custom_fields": {"cache_breakpoint": {}},
+        },
+        # DIAL static tools, resolved by DIAL Core itself
+        {
+            "type": "static_function",
+            "static_function": {"name": "search", "configuration": {}},
+        },
+    ],
+    tool_choice="required",
+    parallel_tool_calls=False,
+    reasoning_effort="high",
+    max_completion_tokens=1000,
+    response_format={
+        "type": "json_schema",
+        "json_schema": {"name": "answer", "schema": {"type": "object"}},
+    },
+    stream_options={"include_usage": True},
+    # DIAL-specific parameters
+    max_prompt_tokens=8000,
+    custom_fields={
+        "configuration": {},
+        "cache_breakpoint": {"expire_at": "5m"},
+    },
+)
+```
+
+The response models cover the DIAL extensions as well:
+
+```pycon
+>>> completion.choices[0].message.custom_content
+CustomContent(
+    stages=[
+        Stage(
+            index=None,
+            name='Thinking',
+            status='completed',
+            content='...',
+            attachments=None
+        )
+    ],
+    attachments=None,
+    state=None,
+    form_value=None,
+    form_schema=None
+)
+>>> completion.usage
+CompletionUsage(
+    prompt_tokens=11,
+    completion_tokens=1,
+    total_tokens=12,
+    prompt_tokens_details=PromptTokensDetails(
+        cached_tokens=8,
+        cache_write_tokens=3
+    ),
+    completion_tokens_details=CompletionTokensDetails(reasoning_tokens=1)
+)
+>>> completion.statistics
+Statistics(
+    usage_per_model=[
+        UsagePerModel(
+            index=0,
+            model='gpt-4o',
+            prompt_tokens=11,
+            completion_tokens=1,
+            total_tokens=12
+        )
+    ],
+    discarded_messages=[0, 1]
+)
+```
+
 ### Working with Files
 
 #### Working with URLs

aidial_client/resources/chat/completions.py

@@ -28,6 +28,10 @@
     FunctionCallSpecParam,
     FunctionParam,
     Message,
+    ReasoningEffort,
+    ResponseFormat,
+    StaticToolParam,
+    StreamOptions,
     ToolCallSpecParam,
     ToolParam,
 )
@@ -51,19 +55,26 @@ def create(
         function_call: Literal["none", "auto"]
         | FunctionCallSpecParam
         | None = None,
-        tools: list[ToolParam] | None = None,
-        tool_choice: Literal["none", "auto"] | ToolCallSpecParam | None = None,
+        tools: list[ToolParam | StaticToolParam] | None = None,
+        tool_choice: Literal["none", "auto", "required"]
+        | ToolCallSpecParam
+        | None = None,
+        parallel_tool_calls: bool | None = None,
         temperature: float | None = None,
         top_p: float | None = None,
         n: int | None = None,
         stop: str | list[str] | None = None,
         max_tokens: int | None = None,
+        max_completion_tokens: int | None = None,
         max_prompt_tokens: Literal["infinity"] | int | None = None,
         presence_penalty: float | None = None,
         frequency_penalty: float | None = None,
         logit_bias: dict | None = None,
         seed: int | None = None,
         user: str | None = None,
+        reasoning_effort: ReasoningEffort | None = None,
+        response_format: ResponseFormat | None = None,
+        stream_options: StreamOptions | None = None,
         custom_fields: ChatCompletionRequestCustomFields | None = None,
         logprobs: bool | None = None,
         top_logprobs: int | None = None,
@@ -86,19 +97,26 @@ def create(
         function_call: Literal["none", "auto"]
         | FunctionCallSpecParam
         | None = None,
-        tools: list[ToolParam] | None = None,
-        tool_choice: Literal["none", "auto"] | ToolCallSpecParam | None = None,
+        tools: list[ToolParam | StaticToolParam] | None = None,
+        tool_choice: Literal["none", "auto", "required"]
+        | ToolCallSpecParam
+        | None = None,
+        parallel_tool_calls: bool | None = None,
         temperature: float | None = None,
         top_p: float | None = None,
         n: int | None = None,
         stop: str | list[str] | None = None,
         max_tokens: int | None = None,
+        max_completion_tokens: int | None = None,
         max_prompt_tokens: Literal["infinity"] | int | None = None,
         presence_penalty: float | None = None,
         frequency_penalty: float | None = None,
         logit_bias: dict | None = None,
         seed: int | None = None,
         user: str | None = None,
+        reasoning_effort: ReasoningEffort | None = None,
+        response_format: ResponseFormat | None = None,
+        stream_options: StreamOptions | None = None,
         custom_fields: ChatCompletionRequestCustomFields | None = None,
         logprobs: bool | None = None,
         top_logprobs: int | None = None,
@@ -120,19 +138,26 @@ def create(
         function_call: Literal["none", "auto"]
         | FunctionCallSpecParam
         | None = None,
-        tools: list[ToolParam] | None = None,
-        tool_choice: Literal["none", "auto"] | ToolCallSpecParam | None = None,
+        tools: list[ToolParam | StaticToolParam] | None = None,
+        tool_choice: Literal["none", "auto", "required"]
+        | ToolCallSpecParam
+        | None = None,
+        parallel_tool_calls: bool | None = None,
         temperature: float | None = None,
         top_p: float | None = None,
         n: int | None = None,
         stop: str | list[str] | None = None,
         max_tokens: int | None = None,
+        max_completion_tokens: int | None = None,
         max_prompt_tokens: Literal["infinity"] | int | None = None,
         presence_penalty: float | None = None,
         frequency_penalty: float | None = None,
         logit_bias: dict | None = None,
         seed: int | None = None,
         user: str | None = None,
+        reasoning_effort: ReasoningEffort | None = None,
+        response_format: ResponseFormat | None = None,
+        stream_options: StreamOptions | None = None,
         custom_fields: ChatCompletionRequestCustomFields | None = None,
         logprobs: bool | None = None,
         top_logprobs: int | None = None,
@@ -165,11 +190,24 @@ def create(
                 "tools": tools,
                 "top_p": top_p,
                 "user": user,
-                "max_prompt_tokens": max_prompt_tokens,
-                "custom_fields": custom_fields,
                 "logprobs": logprobs,
                 "top_logprobs": top_logprobs,
-                "extra_body": extra_body,
+                # DIAL-specific parameters and the ones which aren't supported
+                # by every openai version are sent in the request body directly
+                "extra_body": {
+                    **remove_none(
+                        {
+                            "max_prompt_tokens": max_prompt_tokens,
+                            "custom_fields": custom_fields,
+                            "max_completion_tokens": max_completion_tokens,
+                            "parallel_tool_calls": parallel_tool_calls,
+                            "reasoning_effort": reasoning_effort,
+                            "response_format": response_format,
+                            "stream_options": stream_options,
+                        }
+                    ),
+                    **extra_body,
+                },
                 "extra_query": {
                     "api-version": (
                         api_version or self.default_api_version or Omit()
@@ -217,20 +255,29 @@ async def create(
         function_call: Literal["none", "auto"]
         | FunctionCallSpecParam
         | None = None,
-        tools: list[ToolParam] | None = None,
-        tool_choice: Literal["none", "auto"] | ToolCallSpecParam | None = None,
+        tools: list[ToolParam | StaticToolParam] | None = None,
+        tool_choice: Literal["none", "auto", "required"]
+        | ToolCallSpecParam
+        | None = None,
+        parallel_tool_calls: bool | None = None,
         temperature: float | None = None,
         top_p: float | None = None,
         n: int | None = None,
         stop: str | list[str] | None = None,
         max_tokens: int | None = None,
+        max_completion_tokens: int | None = None,
         max_prompt_tokens: Literal["infinity"] | int | None = None,
         presence_penalty: float | None = None,
         frequency_penalty: float | None = None,
         logit_bias: dict | None = None,
         seed: int | None = None,
         user: str | None = None,
+        reasoning_effort: ReasoningEffort | None = None,
+        response_format: ResponseFormat | None = None,
+        stream_options: StreamOptions | None = None,
         custom_fields: ChatCompletionRequestCustomFields | None = None,
+        logprobs: bool | None = None,
+        top_logprobs: int | None = None,
         # Extra params
         extra_body: dict[str, Any] | None = None,
         extra_headers: Mapping[StrictStr, StrictStr] | None = None,
@@ -250,19 +297,26 @@ async def create(
         function_call: Literal["none", "auto"]
         | FunctionCallSpecParam
         | None = None,
-        tools: list[ToolParam] | None = None,
-        tool_choice: Literal["none", "auto"] | ToolCallSpecParam | None = None,
+        tools: list[ToolParam | StaticToolParam] | None = None,
+        tool_choice: Literal["none", "auto", "required"]
+        | ToolCallSpecParam
+        | None = None,
+        parallel_tool_calls: bool | None = None,
         temperature: float | None = None,
         top_p: float | None = None,
         n: int | None = None,
         stop: str | list[str] | None = None,
         max_tokens: int | None = None,
+        max_completion_tokens: int | None = None,
         max_prompt_tokens: Literal["infinity"] | int | None = None,
         presence_penalty: float | None = None,
         frequency_penalty: float | None = None,
         logit_bias: dict | None = None,
         seed: int | None = None,
         user: str | None = None,
+        reasoning_effort: ReasoningEffort | None = None,
+        response_format: ResponseFormat | None = None,
+        stream_options: StreamOptions | None = None,
         custom_fields: ChatCompletionRequestCustomFields | None = None,
         logprobs: bool | None = None,
         top_logprobs: int | None = None,
@@ -284,19 +338,26 @@ async def create(
         function_call: Literal["none", "auto"]
         | FunctionCallSpecParam
         | None = None,
-        tools: list[ToolParam] | None = None,
-        tool_choice: Literal["none", "auto"] | ToolCallSpecParam | None = None,
+        tools: list[ToolParam | StaticToolParam] | None = None,
+        tool_choice: Literal["none", "auto", "required"]
+        | ToolCallSpecParam
+        | None = None,
+        parallel_tool_calls: bool | None = None,
         temperature: float | None = None,
         top_p: float | None = None,
         n: int | None = None,
         stop: str | list[str] | None = None,
         max_tokens: int | None = None,
+        max_completion_tokens: int | None = None,
         max_prompt_tokens: Literal["infinity"] | int | None = None,
         presence_penalty: float | None = None,
         frequency_penalty: float | None = None,
         logit_bias: dict | None = None,
         seed: int | None = None,
         user: str | None = None,
+        reasoning_effort: ReasoningEffort | None = None,
+        response_format: ResponseFormat | None = None,
+        stream_options: StreamOptions | None = None,
         custom_fields: ChatCompletionRequestCustomFields | None = None,
         logprobs: bool | None = None,
         top_logprobs: int | None = None,
@@ -329,11 +390,24 @@ async def create(
                 "tools": tools,
                 "top_p": top_p,
                 "user": user,
-                "max_prompt_tokens": max_prompt_tokens,
-                "custom_fields": custom_fields,
                 "logprobs": logprobs,
                 "top_logprobs": top_logprobs,
-                "extra_body": extra_body,
+                # DIAL-specific parameters and the ones which aren't supported
+                # by every openai version are sent in the request body directly
+                "extra_body": {
+                    **remove_none(
+                        {
+                            "max_prompt_tokens": max_prompt_tokens,
+                            "custom_fields": custom_fields,
+                            "max_completion_tokens": max_completion_tokens,
+                            "parallel_tool_calls": parallel_tool_calls,
+                            "reasoning_effort": reasoning_effort,
+                            "response_format": response_format,
+                            "stream_options": stream_options,
+                        }
+                    ),
+                    **extra_body,
+                },
                 "extra_query": {
                     "api-version": (
                         api_version or self.default_api_version or Omit()