diff --git a/api/streaming.py b/api/streaming.py
index e752bc8d..7c0c9f2e 100644
--- a/api/streaming.py
+++ b/api/streaming.py
@@ -6694,13 +6694,19 @@ def _run_agent_streaming(
         s = get_session(session_id)
         update_active_run(stream_id, phase="running", session_id=session_id)
         s.workspace = str(Path(workspace).expanduser().resolve())
-        s.model = model
+        _last_persisted_model = getattr(s, "model", None)
         provider_context = (
             str(model_provider).strip().lower()
             if model_provider is not None
             else getattr(s, "model_provider", None)
         )
-        s.model_provider = provider_context or None
+        # #4251: the route layer already persisted this turn's model under the
+        # session lock before dispatch, so a mismatch here means a newer picker
+        # write won the race and must not be clobbered by the worker thread.
+        if _last_persisted_model in (None, "", model):
+            s.model = model
+            s.model_provider = provider_context or None
+            _last_persisted_model = model
 
         _agent_lock = _get_session_agent_lock(session_id)
         # TD1: set thread-local env context so concurrent sessions don't clobber globals
@@ -6742,9 +6748,13 @@ def _run_agent_streaming(
             profile_home=_profile_home,
             has_profile=bool(getattr(s, "profile", None)),
         )
-        s.model_provider = provider_context
-        if _repaired and model != (s.model or ""):
-            s.model = model
+        # #4251: only apply the profile-repair persistence if this turn still
+        # owns the session model value it last wrote.
+        if getattr(s, "model", None) == _last_persisted_model:
+            s.model_provider = provider_context
+            if _repaired and model != (s.model or ""):
+                s.model = model
+                _last_persisted_model = model
 
         # Capture the resolved profile name now, while profile context is
         # reliable. Used in the compression migration block to stamp s.profile
diff --git a/tests/test_issue4251_wakeup_model_picker_race.py b/tests/test_issue4251_wakeup_model_picker_race.py
new file mode 100644
index 00000000..2f2ba5e7
--- /dev/null
+++ b/tests/test_issue4251_wakeup_model_picker_race.py
@@ -0,0 +1,191 @@
+"""Regression tests for wakeup model picker races (#4251)."""
+
+import queue
+import sys
+import types
+from unittest import mock
+
+import api.oauth
+import api.streaming as streaming
+
+
+class FakeSession:
+    def __init__(self, *, model, model_provider="anthropic"):
+        self.session_id = "sess-4251"
+        self.title = "Wakeup picker race"
+        self.workspace = "/tmp"
+        self.model = model
+        self.model_provider = model_provider
+        self.messages = []
+        self.personality = None
+        self.profile = None
+        self.input_tokens = 0
+        self.output_tokens = 0
+        self.estimated_cost = None
+        self.tool_calls = []
+        self.active_stream_id = None
+        self.pending_user_message = None
+        self.pending_attachments = []
+        self.pending_started_at = None
+
+    def save(self, touch_updated_at=True, **kwargs):
+        self._saved = touch_updated_at
+
+    def compact(self):
+        return {
+            "session_id": self.session_id,
+            "title": self.title,
+            "workspace": self.workspace,
+            "model": self.model,
+            "created_at": 0,
+            "updated_at": 0,
+            "pinned": False,
+            "archived": False,
+            "project_id": None,
+            "profile": self.profile,
+            "input_tokens": self.input_tokens,
+            "output_tokens": self.output_tokens,
+            "estimated_cost": self.estimated_cost,
+            "personality": self.personality,
+        }
+
+
+class CapturingAgent:
+    def __init__(
+        self,
+        model=None,
+        provider=None,
+        base_url=None,
+        api_key=None,
+        platform=None,
+        quiet_mode=False,
+        enabled_toolsets=None,
+        fallback_model=None,
+        session_id=None,
+        session_db=None,
+        stream_delta_callback=None,
+        reasoning_callback=None,
+        tool_progress_callback=None,
+        clarify_callback=None,
+        **kwargs,
+    ):
+        self.init_kwargs = {
+            "model": model,
+            "provider": provider,
+            "base_url": base_url,
+            "api_key": api_key,
+            "session_id": session_id,
+            "session_db": session_db,
+        }
+        self.session_id = session_id
+        self.context_compressor = None
+        self.session_prompt_tokens = 0
+        self.session_completion_tokens = 0
+        self.session_estimated_cost_usd = None
+        self.reasoning_config = None
+        self.ephemeral_system_prompt = None
+        self._last_error = None
+
+    def run_conversation(self, **kwargs):
+        return {
+            "messages": [
+                {"role": "user", "content": kwargs["persist_user_message"]},
+                {"role": "assistant", "content": "ok"},
+            ]
+        }
+
+    def interrupt(self, _message):
+        return None
+
+
+def _install_streaming_harness(monkeypatch, fake_session):
+    fake_runtime_module = types.ModuleType("hermes_cli.runtime_provider")
+    fake_runtime_module.resolve_runtime_provider = mock.Mock(
+        return_value={
+            "provider": "anthropic",
+            "base_url": None,
+            "api_key": "rt-key",
+        }
+    )
+    fake_hermes_cli = types.ModuleType("hermes_cli")
+    fake_hermes_cli.runtime_provider = fake_runtime_module
+    fake_hermes_state = types.ModuleType("hermes_state")
+    fake_hermes_state.SessionDB = mock.Mock(return_value=object())
+
+    def fake_runtime_lock(resolver, **kwargs):
+        return resolver(**kwargs)
+
+    monkeypatch.setattr(
+        api.oauth,
+        "resolve_runtime_provider_with_anthropic_env_lock",
+        fake_runtime_lock,
+    )
+    monkeypatch.setattr(streaming, "get_session", lambda _session_id: fake_session)
+    monkeypatch.setattr(streaming, "_get_ai_agent", lambda: CapturingAgent)
+    monkeypatch.setattr(
+        streaming,
+        "resolve_model_provider",
+        lambda *_args, **_kwargs: ("haiku-4-5", "anthropic", None),
+    )
+    monkeypatch.setattr("api.config.get_config", lambda: {})
+    monkeypatch.setattr("api.config._resolve_cli_toolsets", lambda *_args, **_kwargs: [])
+    monkeypatch.setitem(sys.modules, "hermes_cli", fake_hermes_cli)
+    monkeypatch.setitem(sys.modules, "hermes_cli.runtime_provider", fake_runtime_module)
+    monkeypatch.setitem(sys.modules, "hermes_state", fake_hermes_state)
+
+
+def _run_streaming_turn(monkeypatch, fake_session, *, stream_id):
+    _install_streaming_harness(monkeypatch, fake_session)
+    fake_session.active_stream_id = stream_id
+    fake_queue = queue.Queue()
+    try:
+        streaming.STREAMS[stream_id] = fake_queue
+        streaming._run_agent_streaming(
+            session_id=fake_session.session_id,
+            msg_text="background wakeup turn",
+            model="haiku-4-5",
+            model_provider="anthropic",
+            workspace="/tmp",
+            stream_id=stream_id,
+        )
+    finally:
+        streaming.STREAMS.pop(stream_id, None)
+        streaming.AGENT_INSTANCES.pop(stream_id, None)
+
+
+def test_dispatch_stamp_does_not_clobber_newer_picker_model(monkeypatch):
+    fake_session = FakeSession(model="opus-4-8", model_provider="anthropic")
+
+    _run_streaming_turn(
+        monkeypatch,
+        fake_session,
+        stream_id="stream-4251-race",
+    )
+
+    assert fake_session.model == "opus-4-8"
+
+
+def test_dispatch_stamp_persists_resolved_model_when_no_race(monkeypatch):
+    fake_session = FakeSession(model="haiku-4-5", model_provider=None)
+
+    _run_streaming_turn(
+        monkeypatch,
+        fake_session,
+        stream_id="stream-4251-steady",
+    )
+
+    assert fake_session.model == "haiku-4-5"
+    assert fake_session.model_provider == "anthropic"
+
+
+def test_dispatch_stamp_persists_when_session_model_was_empty(monkeypatch):
+    fake_session = FakeSession(model="", model_provider=None)
+
+    _run_streaming_turn(
+        monkeypatch,
+        fake_session,
+        stream_id="stream-4251-empty",
+    )
+
+    assert fake_session.model == "haiku-4-5"
+    assert fake_session.model_provider == "anthropic"