fix(gateway): report the most recent status model

gateway/slash_commands_status.py

@@ -83,7 +83,7 @@ def _quiet_sync(call, default=None):
 def _status_model_route(status_agent, persisted_route: dict, session_row: dict, session_entry):
     """``(model, provider, context_used, context_total)`` for /status.
 
-    Order: live/cached agent route -> persisted dominant route -> SessionDB row -> gateway config
+    Order: live/cached agent route -> persisted recent route -> SessionDB row -> gateway config
     (only loaded when something is still missing).
     """
     from gateway.run import _AGENT_PENDING_SENTINEL, _load_gateway_config, _resolve_gateway_model
@@ -308,7 +308,7 @@ class GatewayStatusCommandsMixin:
             _int_value(session_row.get(k))
             for k in ("input_tokens", "output_tokens", "cache_read_tokens", "cache_write_tokens", "reasoning_tokens")
         )
-        route = await _quiet(lambda: db.get_dominant_session_model_route(session_id))
+        route = await _quiet(lambda: db.get_recent_session_model_route(session_id))
         return title, session_row, db_total_tokens, route if isinstance(route, dict) else {}
 
     @staticmethod

hermes_state_sessions.py

@@ -765,6 +765,23 @@ class SessionSessionsMixin:
         )
         return dict(row) if row else None
 
+    def get_recent_session_model_route(self, session_id: str) -> Optional[Dict[str, Any]]:
+        """Most recently used main-loop model route as one coherent per-call tuple."""
+        self.flush_token_counts()
+        row = self._read_one(
+            """SELECT model, billing_provider, billing_base_url, billing_mode,
+                      api_call_count
+                 FROM session_model_usage
+                WHERE session_id = ?
+                  AND task = ''
+                  AND model <> 'unknown'
+                  AND billing_provider <> ''
+                ORDER BY last_seen DESC
+                LIMIT 1""",
+            (session_id,),
+        )
+        return dict(row) if row else None
+
     def resolve_session_id(self, session_id_or_prefix: str) -> Optional[str]:
         """Exact id, else the single unambiguous prefix match, else None."""
         exact = self.get_session(session_id_or_prefix)

tests/gateway/test_status_command.py

@@ -142,8 +142,8 @@ async def test_status_command_includes_live_agent_model_and_context():
 
 
 @pytest.mark.asyncio
-async def test_status_command_uses_dominant_persisted_model_route(tmp_path):
-    """Persisted status must not combine a model and provider from different calls."""
+async def test_status_command_uses_most_recent_persisted_model_route(tmp_path):
+    """Persisted status uses the latest coherent route, not the lifetime-dominant route."""
     session_entry = SessionEntry(
         session_key=build_session_key(_make_source()),
         session_id="sess-1",
@@ -183,8 +183,8 @@ async def test_status_command_uses_dominant_persisted_model_route(tmp_path):
 
         result = await runner._handle_message(_make_event("/status"))
 
-        assert "**Model:** `z-ai/glm-5.2` (nvidia)" in result
-        assert "**Model:** `z-ai/glm-5.2` (nous)" not in result
+        assert "**Model:** `upstage/solar-pro4:free` (nous)" in result
+        assert "**Model:** `z-ai/glm-5.2` (nvidia)" not in result
     finally:
         db.close()