chore(deps): update python-nonmajor

[renovate-bot]

This PR contains the following updates:

Package	Change	Age	Confidence
aiohttp	==3.13.0 -> ==3.13.1
google-cloud-secret-manager (source)	==2.24.0 -> ==2.25.0
google-cloud-storage	==3.4.0 -> ==3.4.1
langchain-core (source, changelog)	==0.3.78 -> ==0.3.79
llama-index	==0.14.4 -> ==0.14.5
pydantic (changelog)	==2.11.10 -> ==2.12.3

---

Release Notes

aio-libs/aiohttp (aiohttp)

v3.13.1

Compare Source

====================

Bug fixes

• Authentication provided by a redirect now takes precedence over provided auth when making requests with the client -- by :user:PLPeeters.

  Related issues and pull requests on GitHub:
  :issue:9436.

• Fixed :py:meth:WebSocketResponse.close() <aiohttp.web.WebSocketResponse.close> to discard non-close messages within its timeout window after sending close -- by :user:lenard-mosys.

  Related issues and pull requests on GitHub:
  :issue:9506.

• Fixed a deadlock that could occur while attempting to get a new connection slot after a timeout -- by :user:bdraco.

  The connector was not cancellation-safe.

  Related issues and pull requests on GitHub:
  :issue:9670, :issue:9671.

• Fixed the WebSocket flow control calculation undercounting with multi-byte data -- by :user:bdraco.

  Related issues and pull requests on GitHub:
  :issue:9686.

• Fixed incorrect parsing of chunk extensions with the pure Python parser -- by :user:bdraco.

  Related issues and pull requests on GitHub:
  :issue:9851.

• Fixed system routes polluting the middleware cache -- by :user:bdraco.

  Related issues and pull requests on GitHub:
  :issue:9852.

Removals and backward incompatible breaking changes

• Improved performance of the connector when a connection can be reused -- by :user:bdraco.

  If BaseConnector.connect has been subclassed and replaced with custom logic, the ceil_timeout must be added.

  Related issues and pull requests on GitHub:
  :issue:9600.

Miscellaneous internal changes

• Improved performance of the client request lifecycle when there are no cookies -- by :user:bdraco.

  Related issues and pull requests on GitHub:
  :issue:9470.

• Improved performance of sending client requests when the writer can finish synchronously -- by :user:bdraco.

  Related issues and pull requests on GitHub:
  :issue:9485.

• Improved performance of serializing HTTP headers -- by :user:bdraco.

  Related issues and pull requests on GitHub:
  :issue:9603.

• Passing enable_cleanup_closed to :py:class:aiohttp.TCPConnector is now ignored on Python 3.12.7+ and 3.13.1+ since the underlying bug that caused asyncio to leak SSL connections has been fixed upstream -- by :user:bdraco.

  Related issues and pull requests on GitHub:
  :issue:9726, :issue:9736.

---

googleapis/google-cloud-python (google-cloud-secret-manager)

v2.25.0: google-cloud-secret-manager: v2.25.0

Compare Source

Features

• Add support for Python 3.14 (4763aa7)

Bug Fixes

• Deprecate credentials_file argument (4763aa7)

googleapis/python-storage (google-cloud-storage)

v3.4.1

Compare Source

Bug Fixes

• Fixes #​1561 by adding an option to specify the entire object checksum for resumable uploads via the upload_from_string, upload_from_file, and upload_from_filename methods (acb918e)

run-llama/llama_index (llama-index)

v0.14.5

Compare Source

llama-index-core [0.14.5]

• Remove debug print (#​20000)
• safely initialize RefDocInfo in Docstore (#​20031)
• Add progress bar for multiprocess loading (#​20048)
• Fix duplicate node positions when identical text appears multiple times in document (#​20050)
• chore: tool call block - part 1 (#​20074)

llama-index-instrumentation [0.4.2]

• update instrumentation package metadata (#​20079)

llama-index-llms-anthropic [0.9.5]

• ✨ feat(anthropic): add prompt caching model validation utilities (#​20069)
• fix streaming thinking/tool calling with anthropic (#​20077)
• Add haiku 4.5 support (#​20092)

llama-index-llms-baseten [0.1.6]

• Baseten provider Kimi K2 0711, Llama 4 Maverick and Llama 4 Scout Model APIs deprecation (#​20042)

llama-index-llms-bedrock-converse [0.10.5]

• feat: List Claude Sonnet 4.5 as a reasoning model (#​20022)
• feat: Support global cross-region inference profile prefix (#​20064)
• Update utils.py for opus 4.1 (#​20076)
• 4.1 opus bedrockconverse missing in funcitoncalling models (#​20084)
• Add haiku 4.5 support (#​20092)

llama-index-llms-fireworks [0.4.4]

• Add Support for Custom Models in Fireworks LLM (#​20023)
• fix(llms/fireworks): Cannot use Fireworks Deepseek V3.1-20006 issue (#​20028)

llama-index-llms-oci-genai [0.6.3]

• Add support for xAI models in OCI GenAI (#​20089)

llama-index-llms-openai [0.6.4]

• Gpt 5 pro addition (#​20029)
• fix collecting final response with openai responses streaming (#​20037)
• Add support for GPT-5 models in utils.py (JSON_SCHEMA_MODELS) (#​20045)
• chore: tool call block - part 1 (#​20074)

llama-index-llms-sglang [0.1.0]

• Added Sglang llm integration (#​20020)

llama-index-readers-gitlab [0.5.1]

• feat(gitlab): add pagination params for repository tree and issues (#​20052)

llama-index-readers-json [0.4.2]

• vbump the JSON reader (#​20039)

llama-index-readers-web [0.5.5]

• fix: ScrapflyReader Pydantic validation error (#​19999)

llama-index-storage-chat-store-dynamodb [0.4.2]

• bump dynamodb chat store deps (#​20078)

llama-index-tools-mcp [0.4.2]

• 🐛 fix(tools/mcp): Fix dict type handling and reference resolution in … (#​20082)

llama-index-tools-signnow [0.1.0]

• feat(signnow): SignNow mcp tools integration (#​20057)

llama-index-tools-tavily-research [0.4.2]

• feat: Add Tavily extract function for URL content extraction (#​20038)

llama-index-vector-stores-azurepostgresql [0.2.0]

• Add hybrid search to Azure PostgreSQL integration (#​20027)

llama-index-vector-stores-milvus [0.9.3]

• fix: Milvus get_field_kwargs() (#​20086)

llama-index-vector-stores-opensearch [0.6.2]

• fix(opensearch): Correct version check for efficient filtering (#​20067)

llama-index-vector-stores-qdrant [0.8.6]

• fix(qdrant): Allow async-only initialization with hybrid search (#​20005)

pydantic/pydantic (pydantic)

v2.12.3

Compare Source

v2.12.2: 2025-10-14

Compare Source

v2.12.2 (2025-10-14)

What's Changed

Fixes

• Release a new pydantic-core version, as a corrupted CPython 3.10 manylinux2014_aarch64 wheel got uploaded (pydantic-core#1843).
• Fix issue with recursive generic models with a parent model class by @​Viicos in #​12398

Full Changelog: pydantic/pydantic@v2.12.1...v2.12.2

v2.12.1: 2025-10-13

Compare Source

v2.12.1 (2025-10-13)

GitHub release

What's Changed

This is the first 2.12 patch release, addressing most (but not all yet) regressions from the initial 2.12.0 release.

Fixes

• Do not evaluate annotations when inspecting validators and serializers by @​Viicos in #​12355
• Make sure None is converted as NoneType in Python 3.14 by @​Viicos in #​12370
• Backport V1 runtime warning when using Python 3.14 by @​Viicos in #​12367
• Fix error message for invalid validator signatures by @​Viicos in #​12366
• Populate field name in ValidationInfo for validation of default value by @​Viicos in pydantic-core#1826
• Encode credentials in MultiHostUrl builder by @​willswire in pydantic-core#1829
• Respect field serializers when using serialize_as_any serialization flag by @​davidhewitt in pydantic-core#1829
• Fix various RootModel serialization issues by @​davidhewitt in pydantic-core#1836

New Contributors

• @​willswire made their first contribution in pydantic-core#1829

Full Changelog: pydantic/pydantic@v2.12.0...v2.12.1

v2.12.0

Compare Source

GitHub release

What's Changed

This is the final 2.12 release. It features the work of 20 external contributors and provides useful new features, along with initial Python 3.14 support.
Several minor changes (considered non-breaking changes according to our versioning policy)
are also included in this release. Make sure to look into them before upgrading.

Changes (see the alpha and beta releases for additional changes since 2.11):

Packaging

• Update V1 copy to v1.10.24 by @​Viicos in #​12338

New Features

• Add extra parameter to the validate functions by @​anvilpete in #​12233
• Add exclude_computed_fields serialization option by @​Viicos in #​12334
• Add preverse_empty_path URL options by @​Viicos in #​12336
• Add union_format parameter to JSON Schema generation by @​Viicos in #​12147
• Add __qualname__ parameter for create_model by @​Atry in #​12001

Fixes

• Do not try to infer name from lambda definitions in pipelines API by @​Viicos in #​12289
• Use proper namespace for functions in TypeAdapter by @​Viicos in #​12324
• Use Any for context type annotation in TypeAdapter by @​inducer in #​12279
• Expose FieldInfo in pydantic.fields.__all__ by @​Viicos in #​12339
• Respect validation_alias in @validate_call by @​Viicos in #​12340
• Use Any as context annotation in plugin API by @​Viicos in #​12341
• Use proper stacklevel in warnings when possible by @​Viicos in #​12342

New Contributors

• @​anvilpete made their first contribution in #​12233
• @​JonathanWindell made their first contribution in #​12327
• @​inducer made their first contribution in #​12279
• @​Atry made their first contribution in #​12001

---

Configuration

📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

👻 Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.