Skip to content

chore(deps): update dependency openai to 2.15.0 - #324

Open
renovate[bot] wants to merge 1 commit into
mainfrom
renovate/openai-2.x
Open

renovate[bot] wants to merge 1 commit into
mainfrom
renovate/openai-2.x

Conversation

@renovate

@renovate renovate Bot commented Nov 5, 2026

Copy link
Copy Markdown
Contributor

This PR contains the following updates:

Package Change Age Confidence
OpenAI (source) 2.14.0 → 2.15.0 age confidence

Release Notes

openai/openai-dotnet (OpenAI)

v2.15.0

Compare Source

Acknowledgments

Thank you to our developer community members who helped to make the OpenAI client library better with their contributions to this release:

Features Added
  • OpenAI.Chat:
    • Expanded latest-mode OpenTelemetry span attributes for request settings, response metadata, and token-usage details.
    • Added privacy-conscious operation exception logs through the existing client logging configuration. Records contain the exception type without exception messages, stack traces, or request/response content.
    • Added support for new reasoning capabilities.
      • Added the ExtraHigh and Max values to the ChatReasoningEffortLevel extensible enum.
  • OpenAI.Decisions:
    • Added experimental support for the Decisions API, which evaluates text, images, and JSON and returns answers about 10x faster than the Responses API. Get the probability that a condition is true, a choice from a fixed set, or a score against a rubric. Use those answers to classify content, route requests, and prioritize work in your application.
      • Introduced the new DecisionsClient with the following methods: CreateDecision and CreateDecisionAsync.
      • Added support for three types of questions:
        • Predicate questions (DecisionsPredicateQuestion), which return the probability that a condition is true.
        • Choice questions (DecisionsChoiceQuestion), which select from a fixed set of choices.
        • Score questions (DecisionsScoreQuestion), which evaluate input against a rubric of score levels.
      • Added support for user, assistant, developer, and system message items as inputs.
        • User messages (DecisionsUserMessageItem) support input text, images, and JSON.
        • Assistant messages (DecisionsAssistantMessageItem) support output text and refusals.
        • Developer and system messages (DecisionsDeveloperMessageItem and DecisionsSystemMessageItem) support text.
      • Added support for function-call and custom-tool-call history as inputs.
        • Tool outputs support text and images.
        • Supplying tool history does not execute tools.
      • Added support for multimodal question instructions.
      • Added support for prompt caching.
        • Added the PromptCacheKey and PromptCacheOptions properties to DecisionCreationOptions.
        • Added the DecisionsPromptCacheOptions type with the Mode and TimeToLive properties.
        • Added the DecisionsPromptCacheBreakpoint type for explicit breakpoints on supported message or tool-output text and image parts.
  • OpenAI.Responses:
    • Expanded latest-mode OpenTelemetry span attributes for request settings, response metadata, and token-usage details.
    • Added privacy-conscious operation exception logs through the existing client logging configuration. Records contain the exception type without exception messages, stack traces, or request/response content.
    • Added experimental support for OpenTelemetry traces and metrics for non-streaming CreateResponse operations.
      • The instrumentation follows the configured GenAI semantic convention version and does not capture prompts, generated content, instructions, tools, or end-user identifiers.
    • Added experimental support for OpenTelemetry instrumentation of CreateResponseStreamingAsync.
      • Added support for recording terminal usage, errors, cancellation, and early disposal without capturing streaming content.
      • The latest semantic-convention opt-in adds request streaming attributes and first-chunk/output-chunk timing histograms.
    • Added support for specifying the amount of reasoning context preserved across turns.
      • Added the ResponseReasoningContext extensible enum with the Auto, CurrentTurn, and AllTurns values.
      • Added the Context property to ResponseReasoningOptions.
    • Added the protected CreateStreamingResult method to ResponsesClient, which lets derived clients observe or transform the updates of a streaming response while the returned result keeps sole ownership of that response.
    • Added extensibility to the ApplyPatchOperation class hierarchy.
      • Added the ApplyPatchOperationKind extensible enum.
      • Added the Kind property to ApplyPatchOperation.
      • Added a protected internal constructor to ApplyPatchOperation that takes a ApplyPatchOperationKind parameter.
    • Added extensibility to the CodeInterpreterCallOutput class hierarchy.
      • Added the CodeInterpreterCallOutputKind extensible enum.
      • Added the Kind property to CodeInterpreterCallOutput.
      • Added a protected internal constructor to CodeInterpreterCallOutput that takes a CodeInterpreterCallOutputKind parameter.
    • Added extensibility to the CodeInterpreterToolContainerConfiguration class hierarchy.
      • Added the CodeInterpreterToolContainerConfigurationKind extensible enum.
      • Added the Kind property to CodeInterpreterToolContainerConfiguration.
      • Added a protected internal constructor to CodeInterpreterToolContainerConfiguration that takes a CodeInterpreterToolContainerConfigurationKind parameter.
    • Added extensibility to the ComputerCallOutput class hierarchy.
      • Added the ComputerCallOutputKind extensible enum.
      • Added the Kind property to ComputerCallOutput.
      • Added a protected internal constructor to ComputerCallOutput that takes a ComputerCallOutputKind parameter.
    • Added extensibility to the ReasoningSummaryPart class hierarchy.
      • Added the ReasoningSummaryPartKind extensible enum.
      • Added the Kind property to ReasoningSummaryPart.
      • Added a protected internal constructor to ReasoningSummaryPart that takes a ReasoningSummaryPartKind parameter.
    • Added extensibility to the WebSearchAction class hierarchy.
      • Added the WebSearchActionKind extensible enum.
      • Added the Kind property to WebSearchAction.
      • Added a protected internal constructor to WebSearchAction that takes a WebSearchActionKind parameter.
    • Added extensibility to the WebSearchActionSource class hierarchy.
      • Added the WebSearchActionSourceKind extensible enum.
      • Added the Kind property to WebSearchActionSource.
      • Added a protected internal constructor to WebSearchActionSource that takes a WebSearchActionSourceKind parameter.
    • Added extensibility to the WebSearchToolLocation class hierarchy.
      • Added the WebSearchToolLocationKind extensible enum.
      • Added the Kind property to WebSearchToolLocation.
      • Added a protected internal constructor to WebSearchToolLocation that takes a WebSearchToolLocationKind parameter.
Bugs Fixed
  • OpenAI.Assistants:
    • Fixed an issue where the synchronous CreateThreadAndRun protocol method of AssistantClient discarded the caller's RequestOptions. This caused the synchronous CreateThreadAndRun convenience method to ignore its CancellationToken. The asynchronous methods were not affected.
    • Fixed an issue where the CreateThreadAndRun and CreateThreadAndRunStreamingAsync methods of AssistantClient threw a NullReferenceException when the optional threadOptions parameter was omitted.
    • Fixed an issue where the EndIndex property of TextAnnotationUpdate returned the start index for streamed text annotations of a type that the library does not model.
    • Fixed an issue where the ModifyAssistant protocol method of AssistantClient would incorrectly discard its CancellationToken.
  • OpenAI.Chat:
    • Fixed an issue where JsonPatch values for message content and content-part properties were emitted alongside modeled values as duplicate JSON properties instead of taking precedence.
    • Fixed an issue in the serialization of ChatResponseFormat to honor root-level JsonPatch values.
    • Latest-mode operation exception logs now include final API errors and rate limiting.
    • Latest-mode spans and metrics omit cache or reasoning token subsets that exceed their reported totals.
  • OpenAI.Responses:
    • Latest-mode operation exception logs now include final API errors and rate limiting.
    • Latest-mode spans and metrics omit cache or reasoning token subsets that exceed their reported totals.
    • Latest-mode streaming spans now retain observed response metadata across interruption, record abnormal finish reasons and first-chunk latency, and omit the streaming attribute on buffered requests.
Other Changes
  • OpenAI.Chat:
    • Replaced gen_ai.client.token.usage with the current OpenTelemetry inference usage counters and per-operation token histograms when OTEL_SEMCONV_STABILITY_OPT_IN includes gen_ai_latest_experimental. Counters include token modality and available cache and reasoning subsets. The default semantic convention mode remains unchanged.
  • OpenAI.Responses:
    • Replaced gen_ai.client.token.usage with the current OpenTelemetry inference usage counters and per-operation token histograms when OTEL_SEMCONV_STABILITY_OPT_IN includes gen_ai_latest_experimental. Counters include token modality and available cache and reasoning subsets. The default semantic convention mode remains unchanged.
Breaking Changes in Experimental APIs
  • OpenAI.Assistants:
    • Changed the type of the options parameter of the constructors of AssistantClient from OpenAIClientOptions to the functionally-equivalent AssistantClientOptions.
    • The experimental streaming methods are now async-only and return Task<AsyncStreamingResult<T>> instead of AsyncCollectionResult<T>. Awaiting the call sends the request and completes once the service response is established, so its Status, ReasonPhrase, and Headers are available before the updates are read. The result can be enumerated once and owns the response; enumerate it to completion or dispose it to release the connection. The synchronous counterparts have been removed:
      • Removed the CreateRunStreaming, CreateThreadAndRunStreaming, and SubmitToolOutputsToRunStreaming methods of the AssistantClient; use the corresponding *Async methods instead.
  • OpenAI.Audio:
    • The experimental streaming methods are now async-only and return Task<AsyncStreamingResult<T>> instead of AsyncCollectionResult<T>. Awaiting the call sends the request and completes once the service response is established, so its Status, ReasonPhrase, and Headers are available before the updates are read. The result can be enumerated once and owns the response; enumerate it to completion or dispose it to release the connection. The synchronous counterparts have been removed:
      • Removed the GenerateSpeechStreaming and TranscribeAudioStreaming methods of the AudioClient; use the corresponding *Async methods instead.
  • OpenAI.Batch:
    • Changed the type of the options parameter of the constructors of BatchClient from OpenAIClientOptions to the functionally-equivalent BatchClientOptions.
  • OpenAI.Containers:
    • Changed the type of the options parameter of the constructors of ContainerClient from OpenAIClientOptions to the functionally-equivalent ContainerClientOptions.
  • OpenAI.Conversations:
    • Changed the type of the options parameter of the constructors of ConversationClient from OpenAIClientOptions to the functionally-equivalent ConversationClientOptions.
  • OpenAI.Evals:
    • Changed the type of the options parameter of the constructors of EvaluationClient from OpenAIClientOptions to the functionally-equivalent EvaluationClientOptions.
  • OpenAI.FineTuning:
    • Changed the type of the options parameter of the constructors of FineTuningClient from OpenAIClientOptions to the functionally-equivalent FineTuningClientOptions.
  • OpenAI.Graders:
    • Changed the type of the options parameter of the constructors of GraderClient from OpenAIClientOptions to the functionally-equivalent GraderClientOptions.
  • OpenAI.Realtime:
    • The service and session operations on RealtimeClient and RealtimeSessionClient are now async-only. The synchronous service and session operations they previously exposed (session and client-secret creation, sending commands and audio, receiving updates, session configuration, and conversation item helpers) have been removed, along with the underlying synchronous connection path. Use the corresponding *Async APIs instead.
  • OpenAI.Responses:
    • The experimental streaming methods are now async-only and return Task<AsyncStreamingResult<T>> instead of AsyncCollectionResult<T>. Awaiting the call sends the request and completes once the service response is established, so its Status, ReasonPhrase, and Headers are available before the updates are read. The result can be enumerated once and owns the response; enumerate it to completion or dispose it to release the connection. The synchronous counterparts have been removed:
      • Removed the CreateResponseStreaming and GetResponseStreaming methods of the ResponsesClient; use the corresponding *Async methods instead.
  • OpenAI.Skills:
    • Changed the type of the options parameter of the constructors of SkillClient from OpenAIClientOptions to the functionally-equivalent SkillClientOptions.
  • OpenAI.VectorStores:
    • Changed the type of the options parameter of the constructors of VectorStoreClient from OpenAIClientOptions to the functionally-equivalent VectorStoreClientOptions.
  • OpenAI.Videos:
    • Changed the type of the options parameter of the constructors of VideoClient from OpenAIClientOptions to the functionally-equivalent VideoClientOptions.

Configuration

📅 Schedule: (UTC)

  • Branch creation
    • At any time (no schedule defined)
  • Automerge
    • At any time (no schedule defined)

🚦 Automerge: Enabled.

♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.


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

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

@renovate renovate Bot added the dependencies label Nov 5, 2026
@codecov

codecov Bot commented Nov 5, 2026 •

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.29%. Comparing base (10c2bf7) to head (1806c73).

Additional details and impacted files
@@           Coverage Diff           @@
##             main     #324   +/-   ##
=======================================
  Coverage   92.29%   92.29%           
=======================================
  Files         256      256           
  Lines       10927    10927           
  Branches     2389     2389           
=======================================
  Hits        10085    10085           
  Misses        406      406           
  Partials      436      436           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
  • 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

@renovate
renovate Bot force-pushed the renovate/openai-2.x branch 28 times, most recently from 5f34a5d to 2addecd Compare November 10, 2026 09:28
@renovate
renovate Bot force-pushed the renovate/openai-2.x branch 22 times, most recently from 70a6c7d to 4bc30dc Compare November 17, 2026 08:14
@renovate
renovate Bot force-pushed the renovate/openai-2.x branch from 4bc30dc to 1806c73 Compare November 17, 2026 12:24

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

0 participants