Strict mypy check fails on transcription response test

[saxon134]

The project config enables strict mypy, but PYTHONPATH=src python -m mypy src tests fails because client.audio.transcription(...) is typed as APIResponse and the test accesses .text. APIResponse only declares id, object, data, and status, so the checked-in strict type check is broken. Fix by adding a typed transcription response model or declaring the text field on the returned model.

File: tests/test_client.py

Line: 165

Severity: high

Summary: apiresponse text attribute breaks strict mypy

[saxon134]

Coding Loop Plan

Refined Issue

Strict mypy fails because the transcription test accesses .text on the result of client.audio.transcription(...), but AudioResource.transcription is currently typed as returning APIResponse, which does not declare a text attribute.

Implementation Plan

Fix the typing mismatch at the SDK model/resource boundary instead of weakening the test or mypy settings. Add a typed transcription response model with a declared text field, then make both the sync and async transcription resource methods deserialize and return that model. Keep the change limited to the audio transcription response surface. Add or extend tests so client.audio.transcription(...).text and (await client.audio.transcription_async(...)).text are exercised against mocked /v1/audio/transcriptions responses. Verify with strict mypy plus the affected sync and async client tests.

Development Tasks

1. Add a typed transcription response model
   Define a new TranscriptionResponse model in src/globalrouter/_models.py that inherits from GlobalRouterModel and declares the transcription text field, preferably as text: str because the API response and existing test expect it to be present. If this package exports response models through globalrouter.__init__, add TranscriptionResponse to the imports and __all__ there as well.
   Files: src/globalrouter/_models.py, src/globalrouter/init.py
   Tests: python -m mypy src tests

2. Use the transcription model in the audio resource
   Update src/globalrouter/_resources.py to import TranscriptionResponse. Change AudioResource.transcription and AudioResource.transcription_async return annotations from APIResponse to TranscriptionResponse, and pass TranscriptionResponse to request_model / request_model_async for /v1/audio/transcriptions. Do not change unrelated audio speech behavior.
   Files: src/globalrouter/_resources.py
   Tests: python -m mypy src tests, python -m pytest tests/test_client.py::test_native_surface_and_sse_streaming