feat: enable custom LLM support for Crew.test()

This PR enables the Crew.test() method to work with any LLM implementation through the LLM class while maintaining backward compatibility with the openai_model_name parameter.

Changes:
- Added new llm parameter to Crew.test() that accepts string or LLM instance
- Maintained backward compatibility with openai_model_name parameter
- Updated CrewEvaluator to handle any LLM implementation
- Added comprehensive test coverage for both new functionality and backward compatibility

Fixes #2078

Co-Authored-By: Joe Moura <joao@crewai.com>

src/crewai/crew.py

@@ -6,6 +6,8 @@ from concurrent.futures import Future
 from hashlib import md5
 from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
+from crewai.llm import LLM
+
 from pydantic import (
     UUID4,
     BaseModel,
@@ -1076,18 +1078,29 @@ class Crew(BaseModel):
         self,
         n_iterations: int,
         openai_model_name: Optional[str] = None,
+        llm: Optional[Union[str, LLM]] = None,
         inputs: Optional[Dict[str, Any]] = None,
     ) -> None:
-        """Test and evaluate the Crew with the given inputs for n iterations concurrently using concurrent.futures."""
+        """Test and evaluate the Crew with the given inputs for n iterations concurrently using concurrent.futures.
+        
+        Args:
+            n_iterations: Number of test iterations to run
+            openai_model_name: (Deprecated) OpenAI model name to use for evaluation
+            llm: Language model to use for evaluation, can be a string (model name) or LLM instance
+            inputs: Optional dictionary of inputs to pass to the crew
+        """
+        if not (openai_model_name or llm):
+            raise ValueError("Either openai_model_name or llm must be provided")
+
         test_crew = self.copy()
 
         self._test_execution_span = test_crew._telemetry.test_execution_span(
             test_crew,
             n_iterations,
             inputs,
-            openai_model_name,  # type: ignore[arg-type]
-        )  # type: ignore[arg-type]
-        evaluator = CrewEvaluator(test_crew, openai_model_name)  # type: ignore[arg-type]
+            str(llm) if llm else openai_model_name,
+        )
+        evaluator = CrewEvaluator(test_crew, llm or openai_model_name)
 
         for i in range(1, n_iterations + 1):
             evaluator.set_iteration(i)

src/crewai/utilities/evaluators/crew_evaluator_handler.py

@@ -1,4 +1,5 @@
 from collections import defaultdict
+from typing import Union
 
 from pydantic import BaseModel, Field
 from rich.box import HEAVY_EDGE
@@ -6,6 +7,7 @@ from rich.console import Console
 from rich.table import Table
 
 from crewai.agent import Agent
+from crewai.llm import LLM
 from crewai.task import Task
 from crewai.tasks.task_output import TaskOutput
 from crewai.telemetry import Telemetry
@@ -23,7 +25,7 @@ class CrewEvaluator:
 
     Attributes:
         crew (Crew): The crew of agents to evaluate.
-        openai_model_name (str): The model to use for evaluating the performance of the agents (for now ONLY OpenAI accepted).
+        llm (Union[str, LLM]): The language model to use for evaluation. Can be a string (model name) or LLM instance.
         tasks_scores (defaultdict): A dictionary to store the scores of the agents for each task.
         iteration (int): The current iteration of the evaluation.
     """
@@ -32,9 +34,9 @@ class CrewEvaluator:
     run_execution_times: defaultdict = defaultdict(list)
     iteration: int = 0
 
-    def __init__(self, crew, openai_model_name: str):
+    def __init__(self, crew, llm: Union[str, "LLM"]):
         self.crew = crew
-        self.openai_model_name = openai_model_name
+        self.llm = llm if isinstance(llm, LLM) else LLM(model=llm)
         self._telemetry = Telemetry()
         self._setup_for_evaluating()
 
@@ -51,7 +53,7 @@ class CrewEvaluator:
             ),
             backstory="Evaluator agent for crew evaluation with precise capabilities to evaluate the performance of the agents in the crew based on the tasks they have performed",
             verbose=False,
-            llm=self.openai_model_name,
+            llm=self.llm,
         )
 
     def _evaluation_task(
@@ -181,7 +183,7 @@ class CrewEvaluator:
                 self.crew,
                 evaluation_result.pydantic.quality,
                 current_task._execution_time,
-                self.openai_model_name,
+                str(self.llm),
             )
             self.tasks_scores[self.iteration].append(evaluation_result.pydantic.quality)
             self.run_execution_times[self.iteration].append(

tests/crew_test.py

@@ -14,6 +14,7 @@ from crewai.agent import Agent
 from crewai.agents.cache import CacheHandler
 from crewai.crew import Crew
 from crewai.crews.crew_output import CrewOutput
+from crewai.llm import LLM
 from crewai.memory.contextual.contextual_memory import ContextualMemory
 from crewai.process import Process
 from crewai.task import Task
@@ -1123,7 +1124,7 @@ def test_kickoff_for_each_empty_input():
     assert results == []
 
 
-@pytest.mark.vcr(filter_headers=["authorization"])
+@pytest.mark.vcr(filter_headeruvs=["authorization"])
 def test_kickoff_for_each_invalid_input():
     """Tests if kickoff_for_each raises TypeError for invalid input types."""
 
@@ -2814,8 +2815,8 @@ def test_conditional_should_execute():
 @mock.patch("crewai.crew.Crew.kickoff")
 def test_crew_testing_function(kickoff_mock, copy_mock, crew_evaluator):
     task = Task(
-        description="Come up with a list of 5 interesting ideas to explore for an article, then write one amazing paragraph highlight for each idea that showcases how good an article about this topic could be. Return the list of ideas with their paragraph and your notes.",
-        expected_output="5 bullet points with a paragraph for each idea.",
+        description="Test task",
+        expected_output="Expected output",
         agent=researcher,
     )
 
@@ -2844,6 +2845,76 @@ def test_crew_testing_function(kickoff_mock, copy_mock, crew_evaluator):
         ]
     )
 
+@mock.patch("crewai.crew.CrewEvaluator")
+@mock.patch("crewai.crew.Crew.copy")
+@mock.patch("crewai.crew.Crew.kickoff")
+def test_crew_test_with_custom_llm(kickoff_mock, copy_mock, crew_evaluator):
+    task = Task(
+        description="Test task",
+        expected_output="Expected output",
+        agent=researcher,
+    )
+    crew = Crew(agents=[researcher], tasks=[task])
+    custom_llm = LLM(model="gpt-4o-mini")
+
+    copy_mock.return_value = crew
+    crew.test(n_iterations=2, llm=custom_llm, inputs={"topic": "AI"})
+
+    kickoff_mock.assert_has_calls([
+        mock.call(inputs={"topic": "AI"}),
+        mock.call(inputs={"topic": "AI"})
+    ])
+
+    crew_evaluator.assert_has_calls([
+        mock.call(crew, custom_llm),
+        mock.call().set_iteration(1),
+        mock.call().set_iteration(2),
+        mock.call().print_crew_evaluation_result(),
+    ])
+
+@mock.patch("crewai.crew.CrewEvaluator")
+@mock.patch("crewai.crew.Crew.copy")
+@mock.patch("crewai.crew.Crew.kickoff")
+def test_crew_test_with_both_llm_and_model_name(kickoff_mock, copy_mock, crew_evaluator):
+    task = Task(
+        description="Test task",
+        expected_output="Expected output",
+        agent=researcher,
+    )
+    crew = Crew(agents=[researcher], tasks=[task])
+    custom_llm = LLM(model="gpt-4o-mini")
+
+    copy_mock.return_value = crew
+    crew.test(n_iterations=2, llm=custom_llm, openai_model_name="gpt-4", inputs={"topic": "AI"})
+
+    kickoff_mock.assert_has_calls([
+        mock.call(inputs={"topic": "AI"}),
+        mock.call(inputs={"topic": "AI"})
+    ])
+
+    # Should prioritize llm over openai_model_name
+    crew_evaluator.assert_has_calls([
+        mock.call(crew, custom_llm),
+        mock.call().set_iteration(1),
+        mock.call().set_iteration(2),
+        mock.call().print_crew_evaluation_result(),
+    ])
+
+@mock.patch("crewai.crew.CrewEvaluator")
+@mock.patch("crewai.crew.Crew.copy")
+@mock.patch("crewai.crew.Crew.kickoff")
+def test_crew_test_with_no_llm_raises_error(kickoff_mock, copy_mock, crew_evaluator):
+    task = Task(
+        description="Test task",
+        expected_output="Expected output",
+        agent=researcher,
+    )
+    crew = Crew(agents=[researcher], tasks=[task])
+
+    copy_mock.return_value = crew
+    with pytest.raises(ValueError, match="Either openai_model_name or llm must be provided"):
+        crew.test(n_iterations=2, inputs={"topic": "AI"})
+
 
 @pytest.mark.vcr(filter_headers=["authorization"])
 def test_hierarchical_verbose_manager_agent():
@@ -3125,4 +3196,4 @@ def test_multimodal_agent_live_image_analysis():
     # Verify we got a meaningful response
     assert isinstance(result.raw, str)
     assert len(result.raw) > 100  # Expecting a detailed analysis
-    assert "error" not in result.raw.lower()  # No error messages in response
+    assert "error" not in result.raw.lower()  # No error messages in response

tests/utilities/evaluators/test_crew_evaluator_handler.py

@@ -4,6 +4,7 @@ import pytest
 
 from crewai.agent import Agent
 from crewai.crew import Crew
+from crewai.llm import LLM
 from crewai.task import Task
 from crewai.tasks.task_output import TaskOutput
 from crewai.utilities.evaluators.crew_evaluator_handler import (
@@ -23,7 +24,7 @@ class TestCrewEvaluator:
         )
         crew = Crew(agents=[agent], tasks=[task])
 
-        return CrewEvaluator(crew, openai_model_name="gpt-4o-mini")
+        return CrewEvaluator(crew, "gpt-4o-mini")
 
     def test_setup_for_evaluating(self, crew_planner):
         crew_planner._setup_for_evaluating()
@@ -140,3 +141,30 @@ class TestCrewEvaluator:
             execute().pydantic = TaskEvaluationPydanticOutput(quality=9.5)
             crew_planner.evaluate(task_output)
             assert crew_planner.tasks_scores[0] == [9.5]
+
+    def test_crew_evaluator_with_custom_llm(self):
+        agent = Agent(role="Agent 1", goal="Goal 1", backstory="Backstory 1")
+        task = Task(
+            description="Task 1",
+            expected_output="Output 1",
+            agent=agent,
+        )
+        crew = Crew(agents=[agent], tasks=[task])
+        custom_llm = LLM(model="gpt-4o-mini")
+
+        evaluator = CrewEvaluator(crew, custom_llm)
+        assert evaluator.llm == custom_llm
+
+    def test_crew_evaluator_with_model_name(self):
+        agent = Agent(role="Agent 1", goal="Goal 1", backstory="Backstory 1")
+        task = Task(
+            description="Task 1",
+            expected_output="Output 1",
+            agent=agent,
+        )
+        crew = Crew(agents=[agent], tasks=[task])
+        model_name = "gpt-4o-mini"
+
+        evaluator = CrewEvaluator(crew, model_name)
+        assert isinstance(evaluator.llm, LLM)
+        assert evaluator.llm.model == model_name