[Abhijeet Prasad (AbhiPrasad)]

Summary

• Guard Pydantic model classes during interaction materialization so response_format classes do not break traced calls.
• Preserve streamed executable code and code execution results, including their encounter order among text parts per candidate.
• Reduce redundant serialization in tools, embeddings, interaction metadata/results, candidate aggregation, and interaction tool-span cleanup.
• Add mixed text/code-execution stream coverage with new VCR recordings for every supported google-genai version.

Logged shape

Streamed code execution parts now log under executable_code or code_execution_result in provider order. Previously, the pydantic Part attribute check caused these parts to be logged as {"function_call": null}, and aggregation could move code parts after all text.

Validation

• Full test_google_genai(latest) suite: 44 passed, 2 skipped, replay-only.
• The new mixed text/code stream test passed against recordings for latest (2.29.0), 1.75.0, and 1.30.0.
• Focused stream and tool-use tests passed across all three versions.
• Added new cassette files for this test only; existing cassettes were not re-recorded.

Follow-up observation

_reconstruct_interaction_outputs_from_events still materializes each event fragment before merging deltas. Media fragments may therefore create repeated attachment materialization; this optional optimization was left unchanged.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-16T00:31:04.113275Z	9d0d822	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7270897aa3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

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

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

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8fec2fd14f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

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

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

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

[Abhijeet Prasad (AbhiPrasad)]

released with https://github.com/braintrustdata/braintrust-sdk-python/releases/tag/py-sdk-v0.46.1