diff --git a/src/conductor/config/schema.py b/src/conductor/config/schema.py
index e56fe92..ae2a88a 100644
--- a/src/conductor/config/schema.py
+++ b/src/conductor/config/schema.py
@@ -385,6 +385,7 @@ class AgentDef(BaseModel):
       'claude-3-haiku-20240307'
 
     Supports environment variables: ${MODEL:-default_value}
+    Supports Jinja2 templates: {{ workflow.input.model_name }}
     """
 
     input: list[str] = Field(default_factory=list)
diff --git a/src/conductor/executor/agent.py b/src/conductor/executor/agent.py
index 3682703..e68c22d 100644
--- a/src/conductor/executor/agent.py
+++ b/src/conductor/executor/agent.py
@@ -145,6 +145,11 @@ async def execute(
             ProviderError: If agent execution fails.
             ValidationError: If output doesn't match schema or tools are invalid.
         """
+        # Render model field if it contains template expressions
+        if agent.model and ("{{" in agent.model or "{%" in agent.model):
+            rendered_model = self.renderer.render(agent.model, context)
+            agent = agent.model_copy(update={"model": rendered_model})
+
         # Render prompt with context
         rendered_prompt = self.renderer.render(agent.prompt, context)
 
diff --git a/tests/test_executor/test_agent.py b/tests/test_executor/test_agent.py
index bf6c94f..0f65a14 100644
--- a/tests/test_executor/test_agent.py
+++ b/tests/test_executor/test_agent.py
@@ -208,6 +208,75 @@ def test_render_prompt_helper(self, simple_agent: AgentDef) -> None:
         assert "Test question?" in rendered
 
 
+class TestAgentExecutorModelRendering:
+    """Tests for model field template rendering."""
+
+    @pytest.mark.asyncio
+    async def test_model_template_is_rendered(self) -> None:
+        """Test that model field with Jinja2 template is resolved."""
+        agent = AgentDef(
+            name="test",
+            model="{{ workflow.input.selected_model }}",
+            prompt="Do something",
+            output=None,
+        )
+
+        def mock_handler(agent, prompt, context):
+            return {"result": "ok"}
+
+        provider = CopilotProvider(mock_handler=mock_handler)
+        executor = AgentExecutor(provider)
+
+        context = {"workflow": {"input": {"selected_model": "claude-opus-4.6-1m"}}}
+        await executor.execute(agent, context)
+
+        call_history = provider.get_call_history()
+        assert call_history[0]["model"] == "claude-opus-4.6-1m"
+
+    @pytest.mark.asyncio
+    async def test_static_model_is_unchanged(self) -> None:
+        """Test that a static model string passes through unchanged."""
+        agent = AgentDef(
+            name="test",
+            model="gpt-4",
+            prompt="Do something",
+            output=None,
+        )
+
+        def mock_handler(agent, prompt, context):
+            return {"result": "ok"}
+
+        provider = CopilotProvider(mock_handler=mock_handler)
+        executor = AgentExecutor(provider)
+
+        await executor.execute(agent, {})
+
+        call_history = provider.get_call_history()
+        assert call_history[0]["model"] == "gpt-4"
+
+    @pytest.mark.asyncio
+    async def test_model_template_does_not_mutate_original(self) -> None:
+        """Test that rendering model creates a copy, not mutating the original."""
+        agent = AgentDef(
+            name="test",
+            model="{{ workflow.input.model_name }}",
+            prompt="Do something",
+            output=None,
+        )
+
+        def mock_handler(agent, prompt, context):
+            return {"result": "ok"}
+
+        provider = CopilotProvider(mock_handler=mock_handler)
+        executor = AgentExecutor(provider)
+
+        context = {"workflow": {"input": {"model_name": "gpt-5.4"}}}
+        await executor.execute(agent, context)
+
+        # Original agent should still have the template
+        assert agent.model == "{{ workflow.input.model_name }}"
+
+
 class TestAgentExecutorWithTools:
     """Tests for agent execution with tools."""