Transcription response is typed too generically

[saxon134]

AudioResource.transcription() returns APIResponse even though /v1/audio/transcriptions returns a top-level text field. The repository's strict mypy configuration fails on tests/test_client.py:165 with "APIResponse has no attribute text". Add a transcription response model with a text field and use it for both sync and async transcription methods, or otherwise expose text in the declared return type.

File: src/globalrouter/_resources.py

Line: 540

Severity: high

Summary: missing typed transcription response in audioresource

[saxon134]

Coding Loop Plan

Refined Issue

Audio transcription responses are currently modeled as the generic APIResponse even though the endpoint returns a transcription-specific payload with a text field. The SDK should expose a concrete transcription response type and use it for sync and async audio transcription methods.

Implementation Plan

Implementation Plan

Add a narrow transcription response model, wire the audio transcription methods to return it, and cover the behavior with focused tests. Keep the change scoped to response typing for /v1/audio/transcriptions; do not refactor unrelated generic APIResponse usage.

Development Tasks

1. Add a concrete transcription response model
   Define a new Pydantic model, likely named Transcription, in src/globalrouter/_models.py. It should inherit from GlobalRouterModel so unknown provider-specific fields remain allowed, and it should include at least text: str because the existing endpoint fixture returns {"text": "hello"}. Export this model from src/globalrouter/__init__.py by importing it and adding it to __all__.
   Files: src/globalrouter/_models.py, src/globalrouter/init.py
   Tests: python -m mypy src

2. Use the transcription model in audio resource methods
   Update AudioResource.transcription and AudioResource.transcription_async in src/globalrouter/_resources.py to return the new Transcription type instead of APIResponse, and pass Transcription to request_model / request_model_async. Add the new model to the imports from globalrouter._models.
   Files: src/globalrouter/_resources.py
   Tests: python -m mypy src, python -m pytest tests/test_client.py::test_native_surface_and_sse_streaming

3. Add focused tests for transcription typing
   Extend the existing transcription coverage in tests/test_client.py to assert that client.audio.transcription(...) returns an instance of the new Transcription model and exposes .text as "hello". Add async coverage for client.audio.transcription_async(...) if an async audio/native-resource test already exists; otherwise add a small focused async test using httpx.MockTransport for /v1/audio/transcriptions.
   Files: tests/test_client.py
   Tests: python -m pytest tests/test_client.py, python -m mypy src