Add ExtraHigh and Max values to response reasoning effort level

[jairmyree]

This pull request introduces two new reasoning effort levels—ExtraHigh and Max—to the ResponseReasoningEffortLevel type across all supported .NET versions and updates the corresponding type specification. It also updates the test suite and session records to use the new ExtraHigh effort level, reflecting changes in API model versions, test data, and output formats.

Enhancements to Reasoning Effort Levels:

• Added new static members ExtraHigh and Max to the ResponseReasoningEffortLevel struct in OpenAI.Responses for .NET 8.0, .NET 10.0, and .NET Standard 2.0, enabling higher granularity in specifying reasoning effort. [1] [2]
• Updated the type specification in models.tsp to include ExtraHigh ("xhigh") and max as valid values for reasoning effort.

Test and Session Record Updates:

• Modified the ResponsesWithReasoning test to use the new ExtraHigh effort level and the updated model version (gpt-5.6). Also added an assertion to verify the correct effort level is used in the response. [1] [2]
• Updated session records (ResponsesWithReasoning.json and ResponsesWithReasoningAsync.json) to reflect the new effort level (xhigh), model version, and various changes in API request/response structure and metadata, including more detailed tool usage and token usage statistics. [1] [2] [3] [4] [5] [6]

[Copilot]

Pull request overview

This PR extends the Responses reasoning configuration surface by adding two new effort levels (ExtraHigh and Max) to ResponseReasoningEffortLevel, ensuring the SDK and generated/public API listings reflect the updated service contract. It also updates the TypeSpec source-of-truth and refreshes the recorded test artifacts to validate and lock in the new wire values (e.g., "xhigh").

Changes:

• Added ResponseReasoningEffortLevel.ExtraHigh ("xhigh") and ResponseReasoningEffortLevel.Max ("max") to the generated model and API exports for all supported TFMs.
• Updated the Responses TypeSpec (models.tsp) to include "xhigh" and "max" in the allowed reasoning effort values.
• Updated ResponsesWithReasoning test + session recordings to use ExtraHigh and a newer model version, and asserted the echoed effort level.

Reviewed changes

Copilot reviewed 7 out of 17 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tests/SessionRecords/ResponsesTests/ResponsesWithReasoningAsync.json	Updated async recording to use "xhigh" reasoning effort and refreshed request/response payload shape/metadata.
tests/SessionRecords/ResponsesTests/ResponsesWithReasoning.json	Updated sync recording to use "xhigh" reasoning effort and refreshed request/response payload shape/metadata.
tests/Responses/ResponsesTests.cs	Switched test to ResponseReasoningEffortLevel.ExtraHigh, updated model id, and added an assertion for the returned effort level.
specification/base/typespec/responses/models.tsp	Extended ResponsesReasoningEffort union with ExtraHigh: "xhigh" and "max".
OpenAI.Responses/src/Generated/Models/ResponseReasoningEffortLevel.cs	Added constants + static members for ExtraHigh and Max in the generated effort level wrapper.
api/netstandard2.0/OpenAI.Responses.netstandard2.0.cs	Updated public API listing to include ExtraHigh and Max.
api/net8.0/OpenAI.Responses.net8.0.cs	Updated public API listing to include ExtraHigh and Max.
api/net10.0/OpenAI.Responses.net10.0.cs	Updated public API listing to include ExtraHigh and Max.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 8 out of 9 changed files in this pull request and generated no new comments.

Suppressed comments (1)

specification/base/typespec/responses/models.tsp:18

• The ResponsesReasoningEffort docstring still lists supported values up through high, but the union now also includes xhigh and max. This makes the TypeSpec documentation (and any generated docs derived from it) misleading.

  Currently supported values are `none`, `minimal`, `low`, `medium`, and `high`. Reducing
  reasoning effort can result in faster responses and fewer tokens used
  on reasoning in a response.
  """)
union ResponsesReasoningEffort {

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 958a8e112f

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[Copilot]

Pull request overview

Copilot reviewed 11 out of 12 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 8 out of 9 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 8 out of 9 changed files in this pull request and generated no new comments.