fix(openai): preserve audio in Responses input

[Mark Xian (xianjianlf2)]

Summary

• convert v1 audio/wav and audio/mp3 blocks to Responses API input_audio items
• preserve Uint8Array audio by encoding it as base64
• reject unsupported audio MIME types explicitly rather than silently dropping the attachment
• add regression coverage for a legacy HumanMessage audio block

Validation

• git diff --check

The focused Vitest suite could not start in this environment because the concurrent pnpm workspace install did not create the required internal package links; the failure was module resolution for @langchain/tsconfig, before the test file executed.

Fixes #11460

[changeset-bot]

🦋 Changeset detected

Latest commit: e8a0f05

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 8 packages

Name	Type
@langchain/openai	Patch
@langchain/classic	Patch
@langchain/deepseek	Patch
@langchain/fireworks	Patch
@langchain/openrouter	Patch
@langchain/together-ai	Patch
@langchain/xai	Patch
@langchain/neo4j	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR