diff --git a/ai-hub/integration_tests/API_COVERAGE.md b/ai-hub/integration_tests/API_COVERAGE.md
index 5ce45d4..61938df 100644
--- a/ai-hub/integration_tests/API_COVERAGE.md
+++ b/ai-hub/integration_tests/API_COVERAGE.md
@@ -4,9 +4,9 @@
 
 ## Todo Works
 - [ ] Parse mapping of remaining APIs from all markdown files in `docs/api_reference/` to complete the checklist.
-- [ ] Build a generic `test_stt_tts.py` integration spec (Targeting `/audio` transcode routing or native gemini).
+- [x] Build a generic `test_audio.py` integration spec (Targeting STT and TTS routing).
 - [ ] Build a `test_agent_nodes_sandbox.py` spec for deep sandbox edge cases to ensure `workspace` restrictions work.
-- [ ] Fully map coverage for the `/models` endpoints to `test_provider_config.py`.
+- [x] Fully map coverage for the `/models` endpoints to `test_provider_config.py`.
 - [ ] Write missing tests for `/users/me/profile` endpoints.
 - [ ] Review `/skills` and `/knowledge` mapping against `test_browser_llm.py` which only touches standard capability mapping.
 
@@ -58,4 +58,11 @@
 Covered roughly by `integration_tests/test_provider_config.py`
 - [x] `POST /users/me/config/verify_llm` (Config validation proxy ping before save)
 - [x] `PUT /users/me/config` (Apply Provider Key Update)
-- [ ] Check full coverage against `api_reference/models.md`
+- [x] `GET /users/me/config` (Get Personal Setting Mapping)
+- [x] `GET /users/me/config/models` (Get Models for Provider)
+
+### Audio Pipeline (STT & TTS)
+Covered by `integration_tests/test_audio.py`
+- [x] `GET /speech/voices` (List available TTS voices)
+- [x] `POST /speech` (Generate speech from text)
+- [x] `POST /stt/transcribe` (Transcribe audio to text)
diff --git a/ai-hub/integration_tests/test_audio.py b/ai-hub/integration_tests/test_audio.py
new file mode 100644
index 0000000..68474ff
--- /dev/null
+++ b/ai-hub/integration_tests/test_audio.py
@@ -0,0 +1,64 @@
+import os
+import httpx
+import pytest
+
+BASE_URL = os.getenv("SYNC_TEST_BASE_URL", "http://127.0.0.1:8002/api/v1")
+
+def _headers():
+    return {
+        "X-User-ID": os.environ.get("SYNC_TEST_USER_ID", "")
+    }
+
+def test_tts_voices():
+    """Test retrieving available TTS voices."""
+    with httpx.Client(timeout=10.0) as client:
+        r = client.get(f"{BASE_URL}/speech/voices", headers=_headers())
+        assert r.status_code == 200, f"Failed to get voices: {r.text}"
+        voices = r.json()
+        assert isinstance(voices, list), "Voices should be a list"
+
+def test_tts_to_stt_lifecycle():
+    """
+    Test generating speech from text (TTS), then transcribing that audio 
+    back to text (STT) to verify the full audio processing pipeline.
+    """
+    user_id = os.environ.get("SYNC_TEST_USER_ID", "")
+    assert user_id, "User ID not found in environment from conftest."
+    
+    test_phrase = "Hello from integration test audio pipeline."
+
+    with httpx.Client(timeout=30.0) as client:
+        # Step 1: Generate speech (TTS)
+        tts_payload = {
+            "text": test_phrase
+        }
+        r_tts = client.post(
+            f"{BASE_URL}/speech", 
+            params={"stream": False}, 
+            headers=_headers(), 
+            json=tts_payload
+        )
+        assert r_tts.status_code == 200, f"TTS failed: {r_tts.text}"
+        
+        # Ensure we got audio bytes back
+        audio_content = r_tts.content
+        assert len(audio_content) > 1000, "TTS audio content seems too small"
+        
+        # Step 2: Transcribe the generated audio (STT)
+        files = {
+            "audio_file": ("test_audio_pipeline.wav", audio_content, "audio/wav")
+        }
+        r_stt = client.post(
+            f"{BASE_URL}/stt/transcribe", 
+            headers=_headers(),
+            files=files
+        )
+        assert r_stt.status_code == 200, f"STT failed: {r_stt.text}"
+        
+        stt_result = r_stt.json()
+        transcript = stt_result.get("transcript", "").lower()
+        
+        # Assert the transcript contains our original phrase (or at least parts of it)
+        # Using a looser verification because STT models might slightly rephrase or omit punctuation
+        assert "hello" in transcript, f"Expected 'hello' in transcript: {transcript}"
+        assert "integration" in transcript, f"Expected 'integration' in transcript: {transcript}"
diff --git a/ai-hub/integration_tests/test_provider_config.py b/ai-hub/integration_tests/test_provider_config.py
index bc54cff..5c84d6e 100644
--- a/ai-hub/integration_tests/test_provider_config.py
+++ b/ai-hub/integration_tests/test_provider_config.py
@@ -124,3 +124,21 @@
         data = r.json()
         # Should gracefully fail
         assert data["success"] is False
+
+def test_get_provider_models():
+    """
+    Test 6: Check retrieving the list of models for a provider via GET /users/me/config/models
+    """
+    with httpx.Client(timeout=10.0) as client:
+        r = client.get(f"{BASE_URL}/users/me/config/models", params={"provider_name": "gemini"}, headers=_headers())
+        assert r.status_code == 200, f"Expected 200 OK: {r.text}"
+        
+        models = r.json()
+        assert isinstance(models, list), "Response should be a list"
+        
+        # We expect at least some gemini models to be returned 
+        # (e.g. gemini-3-flash-preview, gemini-2.5-flash, etc.)
+        assert len(models) > 0, "No models returned for provider 'gemini'"
+        
+        # Verify the structure of the returned objects
+        assert "model_name" in models[0], "Missing 'model_name' key in model object"