diff --git a/api/.env.example b/api/.env.example
index c7df8f0..0d6b52a 100644
--- a/api/.env.example
+++ b/api/.env.example
@@ -9,3 +9,10 @@ REPLICATE_API_TOKEN=
 
 # Port the proxy binds. compose / fly.io override this.
 PORT=8000
+
+# Narrator model (charter §5 "the good model"). qwen3.x runs with thinking OFF.
+OLLAMA_NARRATOR_MODEL=qwen3.5:latest
+# Per-attempt model-call timeout (seconds).
+OLLAMA_TIMEOUT_SECONDS=30
+# Optional: append call logs (JSON lines, §10) to this file instead of stdout.
+CALL_LOG_PATH=
diff --git a/api/conftest.py b/api/conftest.py
new file mode 100644
index 0000000..38d7def
--- /dev/null
+++ b/api/conftest.py
@@ -0,0 +1,21 @@
+"""Gate `@pytest.mark.live` tests (they hit a real Ollama) behind --run-live so
+the default suite stays hermetic. Run the live layer with: pytest --run-live
+"""
+
+import pytest
+
+
+def pytest_addoption(parser):
+    parser.addoption(
+        "--run-live", action="store_true", default=False,
+        help="run tests marked @pytest.mark.live (require a real Ollama)",
+    )
+
+
+def pytest_collection_modifyitems(config, items):
+    if config.getoption("--run-live"):
+        return
+    skip_live = pytest.mark.skip(reason="needs --run-live (real Ollama)")
+    for item in items:
+        if "live" in item.keywords:
+            item.add_marker(skip_live)
diff --git a/api/pytest.ini b/api/pytest.ini
index c038626..795b6eb 100644
--- a/api/pytest.ini
+++ b/api/pytest.ini
@@ -3,3 +3,5 @@ pythonpath = .
 testpaths = tests
 filterwarnings =
     ignore:Using `httpx` with `starlette.testclient` is deprecated
+markers =
+    live: exercises a real Ollama; skipped unless --run-live
diff --git a/api/tests/test_live_smoke.py b/api/tests/test_live_smoke.py
new file mode 100644
index 0000000..c76066c
--- /dev/null
+++ b/api/tests/test_live_smoke.py
@@ -0,0 +1,19 @@
+import json
+from pathlib import Path
+
+import pytest
+
+import app.narrate as narrate
+
+VALID = json.loads((Path(__file__).parent / "fixtures" / "canon_log_valid.json").read_text())
+
+
+@pytest.mark.live
+def test_narrator_returns_real_prose():
+    """Hits the real Ollama (OLLAMA_BASE_URL). Proves the wire end-to-end:
+    prompt assembly → model → non-empty prose. Not asserted: tag content
+    (nondeterministic). Run with: pytest --run-live
+    """
+    text = narrate.run(VALID)
+    assert isinstance(text, str)
+    assert len(text.strip()) > 40  # real prose, not a blip