release: 0.34.0

.release-please-manifest.json

@@ -1,3 +1,3 @@
 {
-  ".": "0.33.0"
+  ".": "0.34.0"
 }

.stats.yml

@@ -1,4 +1,4 @@
 configured_endpoints: 17
-openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/groqcloud%2Fgroqcloud-498dee72535c9b0585491899c8e5173ac2b83676710c741ef41faef415821683.yml
-openapi_spec_hash: aff11e3516f652d05f5416f6593994d1
+openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/groqcloud%2Fgroqcloud-5d2697ad30ccbbd1502ade612deb9221a1e05e70d09a156da743d0ca120942f9.yml
+openapi_spec_hash: 0bde05b560fbd40711e91e6b562c3b0f
 config_hash: a4a6c3089a2e53425351cc9f42b4b5aa

CHANGELOG.md

@@ -1,5 +1,24 @@
 # Changelog
 
+## 0.34.0 (2025-11-05)
+
+Full Changelog: [v0.33.0...v0.34.0](https://github.com/groq/groq-python/compare/v0.33.0...v0.34.0)
+
+### Features
+
+* **api:** api update ([51bb951](https://github.com/groq/groq-python/commit/51bb951f04069bdbc79fbcc96e9a89ad9bb945ed))
+
+
+### Bug Fixes
+
+* **client:** close streams without requiring full consumption ([99b9431](https://github.com/groq/groq-python/commit/99b9431a9663dc1c2abb8f46c2444fa2bb397eb4))
+
+
+### Chores
+
+* **internal/tests:** avoid race condition with implicit client cleanup ([3af43e5](https://github.com/groq/groq-python/commit/3af43e5855668c3632a9053222d11c27fa40c70c))
+* **internal:** grammar fix (it's -> its) ([37687bf](https://github.com/groq/groq-python/commit/37687bfc136e87b85f26f9617b02a80e21047858))
+
 ## 0.33.0 (2025-10-21)
 
 Full Changelog: [v0.32.0...v0.33.0](https://github.com/groq/groq-python/compare/v0.32.0...v0.33.0)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "groq"
-version = "0.33.0"
+version = "0.34.0"
 description = "The official Python library for the groq API"
 dynamic = ["readme"]
 license = "Apache-2.0"

src/groq/_streaming.py

@@ -96,9 +96,8 @@ def __stream__(self) -> Iterator[_T]:
 
                 yield process_data(data={"data": data, "event": sse.event}, cast_to=cast_to, response=response)
 
-        # Ensure the entire stream is consumed
-        for _sse in iterator:
-            ...
+        # As we might not fully consume the response stream, we need to close it explicitly
+        response.close()
 
     def __enter__(self) -> Self:
         return self
@@ -198,9 +197,8 @@ async def __stream__(self) -> AsyncIterator[_T]:
 
                 yield process_data(data={"data": data, "event": sse.event}, cast_to=cast_to, response=response)
 
-        # Ensure the entire stream is consumed
-        async for _sse in iterator:
-            ...
+        # As we might not fully consume the response stream, we need to close it explicitly
+        await response.aclose()
 
     async def __aenter__(self) -> Self:
         return self

src/groq/_utils/_utils.py

@@ -133,7 +133,7 @@ def is_given(obj: _T | NotGiven | Omit) -> TypeGuard[_T]:
 # Type safe methods for narrowing types with TypeVars.
 # The default narrowing for isinstance(obj, dict) is dict[unknown, unknown],
 # however this cause Pyright to rightfully report errors. As we know we don't
-# care about the contained types we can safely use `object` in it's place.
+# care about the contained types we can safely use `object` in its place.
 #
 # There are two separate functions defined, `is_*` and `is_*_t` for different use cases.
 # `is_*` is for when you're dealing with an unknown input

src/groq/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "groq"
-__version__ = "0.33.0"  # x-release-please-version
+__version__ = "0.34.0"  # x-release-please-version

src/groq/types/chat/chat_completion_message.py

@@ -8,6 +8,9 @@
 
 __all__ = [
     "ChatCompletionMessage",
+    "Annotation",
+    "AnnotationDocumentCitation",
+    "AnnotationFunctionCitation",
     "ExecutedTool",
     "ExecutedToolBrowserResult",
     "ExecutedToolCodeResult",
@@ -19,6 +22,45 @@
 ]
 
 
+class AnnotationDocumentCitation(BaseModel):
+    document_id: str
+    """
+    The ID of the document being cited, corresponding to a document provided in the
+    request.
+    """
+
+    end_index: int
+    """The character index in the message content where this citation ends."""
+
+    start_index: int
+    """The character index in the message content where this citation begins."""
+
+
+class AnnotationFunctionCitation(BaseModel):
+    end_index: int
+    """The character index in the message content where this citation ends."""
+
+    start_index: int
+    """The character index in the message content where this citation begins."""
+
+    tool_call_id: str
+    """
+    The ID of the tool call being cited, corresponding to a tool call made during
+    the conversation.
+    """
+
+
+class Annotation(BaseModel):
+    type: Literal["document_citation", "function_citation"]
+    """The type of annotation."""
+
+    document_citation: Optional[AnnotationDocumentCitation] = None
+    """A citation referencing a specific document that was provided in the request."""
+
+    function_citation: Optional[AnnotationFunctionCitation] = None
+    """A citation referencing the result of a function or tool call."""
+
+
 class ExecutedToolBrowserResult(BaseModel):
     title: str
     """The title of the browser window"""
@@ -189,6 +231,12 @@ class ChatCompletionMessage(BaseModel):
     role: Literal["assistant"]
     """The role of the author of this message."""
 
+    annotations: Optional[List[Annotation]] = None
+    """
+    A list of annotations providing citations and references for the content in the
+    message.
+    """
+
     executed_tools: Optional[List[ExecutedTool]] = None
     """
     A list of tools that were executed during the chat completion for compound AI