From aa4ace614e92fea581f55e1730ace9dcafb10fac Mon Sep 17 00:00:00 2001
From: David del Real Sifuentes <deldavid@google.com>
Date: Thu, 16 Jul 2026 22:40:56 +0000
Subject: [PATCH 1/4]  refactor(generative_ai/evaluation): migrate
 vertexai.generative_models to genai SDK

 - Migrated evaluation samples to use the new `google.genai` SDK instead of `vertexai.generative_models`.
 - Removed unnecessary dependencies from requirements after testing confirmed they are no longer needed.
---
 generative_ai/evaluation/get_rouge_score.py   | 12 ++--
 generative_ai/evaluation/noxfile_config.py    |  2 +-
 .../pairwise_summarization_quality.py         | 68 ++++++++++++-------
 .../evaluation/requirements-test.txt          |  6 +-
 generative_ai/evaluation/requirements.txt     | 17 ++---
 5 files changed, 59 insertions(+), 46 deletions(-)

diff --git a/generative_ai/evaluation/get_rouge_score.py b/generative_ai/evaluation/get_rouge_score.py
index 579c0931374..323750f95d6 100644
--- a/generative_ai/evaluation/get_rouge_score.py
+++ b/generative_ai/evaluation/get_rouge_score.py
@@ -11,19 +11,21 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
+# [START generativeaionvertexai_evaluation_get_rouge_score]
+
 import os
 
+import pandas as pd
+
+import vertexai
 from vertexai.preview.evaluation import EvalResult
+from vertexai.preview.evaluation import EvalTask
 
 PROJECT_ID = os.getenv("GOOGLE_CLOUD_PROJECT")
 
 
 def get_rouge_score() -> EvalResult:
-    # [START generativeaionvertexai_evaluation_get_rouge_score]
-    import pandas as pd
-
-    import vertexai
-    from vertexai.preview.evaluation import EvalTask
 
     # TODO(developer): Update & uncomment line below
     # PROJECT_ID = "your-project-id"
diff --git a/generative_ai/evaluation/noxfile_config.py b/generative_ai/evaluation/noxfile_config.py
index 0973c8621c7..1d2019a0b18 100644
--- a/generative_ai/evaluation/noxfile_config.py
+++ b/generative_ai/evaluation/noxfile_config.py
@@ -22,7 +22,7 @@
 
 TEST_CONFIG_OVERRIDE = {
     # You can opt out from the test for specific Python versions.
-    "ignored_versions": ["3.8", "3.9", "3.11", "3.12", "3.13"],
+    "ignored_versions": ["3.8", "3.9", "3.10", "3.12", "3.13"],
     # Old samples are opted out of enforcing Python type hints
     # All new samples should feature them
     "enforce_type_hints": True,
diff --git a/generative_ai/evaluation/pairwise_summarization_quality.py b/generative_ai/evaluation/pairwise_summarization_quality.py
index 88c89871904..ea116a5cb8d 100644
--- a/generative_ai/evaluation/pairwise_summarization_quality.py
+++ b/generative_ai/evaluation/pairwise_summarization_quality.py
@@ -11,28 +11,50 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
+# [START generativeaionvertexai_evaluation_pairwise_summarization_quality]
+
 import os
 
-from vertexai.preview.evaluation import EvalResult
+import pandas as pd
 
+from google import genai
+from google.genai import types
+
+import vertexai
+from vertexai.preview.evaluation import EvalResult
+from vertexai.evaluation import (
+    EvalTask,
+    PairwiseMetric,
+    MetricPromptTemplateExamples,
+)
+
+# TODO (developer) set GOOGLE_CLOUD_PROJECT and REGION_ID
+# environment variables before running.
 PROJECT_ID = os.getenv("GOOGLE_CLOUD_PROJECT")
+LOCATION = os.getenv("REGION_ID")
+BASELINE_MODEL = "gemini-2.5-flash"
 
 
-def evaluate_output() -> EvalResult:
-    # [START generativeaionvertexai_evaluation_pairwise_summarization_quality]
-    import pandas as pd
-
-    import vertexai
-    from vertexai.generative_models import GenerativeModel
-    from vertexai.evaluation import (
-        EvalTask,
-        PairwiseMetric,
-        MetricPromptTemplateExamples,
+def custom_model_fn(prompt: str) -> str:
+    """Generates text from a prompt using the baseline Gemini model via Vertex AI."""
+
+    genai_client = genai.Client(vertexai=True, project=PROJECT_ID, location=LOCATION)
+
+    genai_config = types.GenerateContentConfig(temperature=0.4)
+
+    response = genai_client.models.generate_content(
+        model=BASELINE_MODEL, contents=prompt, config=genai_config
     )
 
-    # TODO(developer): Update & uncomment line below
-    # PROJECT_ID = "your-project-id"
-    vertexai.init(project=PROJECT_ID, location="us-central1")
+    return response.text
+
+
+def evaluate_output() -> EvalResult:
+    """
+    Evaluates a candidate model's summarization quality
+    against a baseline model using Vertex AI.
+    """
 
     prompt = """
     Summarize the text such that a five-year-old can understand.
@@ -49,15 +71,9 @@ def evaluate_output() -> EvalResult:
     efficient, environmentally conscious urban transportation.
     """
 
-    eval_dataset = pd.DataFrame({"prompt": [prompt]})
-
-    # Baseline model for pairwise comparison
-    baseline_model = GenerativeModel("gemini-2.0-flash-lite-001")
+    vertexai.init(project=PROJECT_ID, location=LOCATION)
 
-    # Candidate model for pairwise comparison
-    candidate_model = GenerativeModel(
-        "gemini-2.0-flash-001", generation_config={"temperature": 0.4}
-    )
+    eval_dataset = pd.DataFrame({"prompt": [prompt]})
 
     prompt_template = MetricPromptTemplateExamples.get_prompt_template(
         "pairwise_summarization_quality"
@@ -66,7 +82,7 @@ def evaluate_output() -> EvalResult:
     summarization_quality_metric = PairwiseMetric(
         metric="pairwise_summarization_quality",
         metric_prompt_template=prompt_template,
-        baseline_model=baseline_model,
+        baseline_model=BASELINE_MODEL,
     )
 
     eval_task = EvalTask(
@@ -74,13 +90,17 @@ def evaluate_output() -> EvalResult:
         metrics=[summarization_quality_metric],
         experiment="pairwise-experiment",
     )
-    result = eval_task.evaluate(model=candidate_model)
+    result = eval_task.evaluate(
+        model=custom_model_fn, experiment_run_name="genai-client-vs-legacy-baseline-6"
+    )
 
     baseline_model_response = result.metrics_table["baseline_model_response"].iloc[0]
     candidate_model_response = result.metrics_table["response"].iloc[0]
+
     winner_model = result.metrics_table[
         "pairwise_summarization_quality/pairwise_choice"
     ].iloc[0]
+
     explanation = result.metrics_table[
         "pairwise_summarization_quality/explanation"
     ].iloc[0]
diff --git a/generative_ai/evaluation/requirements-test.txt b/generative_ai/evaluation/requirements-test.txt
index baa23bf9c3e..598922c8c25 100644
--- a/generative_ai/evaluation/requirements-test.txt
+++ b/generative_ai/evaluation/requirements-test.txt
@@ -1,4 +1,4 @@
 backoff==2.2.1
-google-api-core==2.19.0
-pytest==9.0.3; python_version >= "3.10"
-pytest-asyncio==0.23.6
+google-api-core==2.31.0
+pytest==9.0.3
+pytest-asyncio==1.4.0
diff --git a/generative_ai/evaluation/requirements.txt b/generative_ai/evaluation/requirements.txt
index 18ff8773d69..842e7cc88b6 100644
--- a/generative_ai/evaluation/requirements.txt
+++ b/generative_ai/evaluation/requirements.txt
@@ -1,13 +1,4 @@
-pandas==2.2.3; python_version == '3.7'
-pandas==2.2.3; python_version == '3.8'
-pandas==2.2.3; python_version > '3.8'
-pillow==12.2.0
-google-cloud-aiplatform[full]==1.157.0
-sentencepiece==0.2.1
-google-auth==2.38.0
-anthropic[vertex]==0.28.0
-langchain-core==1.4.0
-langchain-google-vertexai==3.2.3
-numpy<3
-openai==1.68.2
-immutabledict==4.2.0
+pandas==2.3.3
+google-auth==2.55.2
+google-cloud-aiplatform[full]==1.161.0
+google-genai==2.12.1

From f55a0531b89f2b2937f8bcbe209d9fbf7907054b Mon Sep 17 00:00:00 2001
From: David del Real Sifuentes <deldavid@google.com>
Date: Mon, 20 Jul 2026 20:35:29 +0000
Subject: [PATCH 2/4] - Migrated sample to new folder under genai. - Refactored
 sample to use newer genai SDK. - Cleaned requirements files from unneccessary
 dependencies.

---
 genai/evaluation/get_rouge_score.py           | 101 +++++++++++++++
 genai/evaluation/noxfile_config.py            |  42 +++++++
 .../pairwise_summarization_quality.py         | 119 ++++++++++++++++++
 genai/evaluation/requirements-test.txt        |   1 +
 genai/evaluation/requirements.txt             |   4 +
 genai/evaluation/test_evaluation.py           |  26 ++++
 generative_ai/evaluation/get_rouge_score.py   |  14 +--
 generative_ai/evaluation/noxfile_config.py    |   2 +-
 .../pairwise_summarization_quality.py         |  68 ++++------
 .../evaluation/requirements-test.txt          |   6 +-
 generative_ai/evaluation/requirements.txt     |  17 ++-
 11 files changed, 340 insertions(+), 60 deletions(-)
 create mode 100644 genai/evaluation/get_rouge_score.py
 create mode 100644 genai/evaluation/noxfile_config.py
 create mode 100644 genai/evaluation/pairwise_summarization_quality.py
 create mode 100644 genai/evaluation/requirements-test.txt
 create mode 100644 genai/evaluation/requirements.txt
 create mode 100644 genai/evaluation/test_evaluation.py

diff --git a/genai/evaluation/get_rouge_score.py b/genai/evaluation/get_rouge_score.py
new file mode 100644
index 00000000000..20e5c7871ff
--- /dev/null
+++ b/genai/evaluation/get_rouge_score.py
@@ -0,0 +1,101 @@
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# TODO: to be refactored in b/536110099
+
+import os
+
+import pandas as pd
+
+import vertexai
+from vertexai.preview.evaluation import EvalResult
+from vertexai.preview.evaluation import EvalTask
+
+PROJECT_ID = os.getenv("GOOGLE_CLOUD_PROJECT")
+
+
+def get_rouge_score() -> EvalResult:
+
+    # TODO(developer): Update & uncomment line below
+    # PROJECT_ID = "your-project-id"
+    vertexai.init(project=PROJECT_ID, location="us-central1")
+
+    reference_summarization = """
+    The Great Barrier Reef, the world's largest coral reef system, is
+    located off the coast of Queensland, Australia. It's a vast
+    ecosystem spanning over 2,300 kilometers with thousands of reefs
+    and islands. While it harbors an incredible diversity of marine
+    life, including endangered species, it faces serious threats from
+    climate change, ocean acidification, and coral bleaching."""
+
+    # Compare pre-generated model responses against the reference (ground truth).
+    eval_dataset = pd.DataFrame(
+        {
+            "response": [
+                """The Great Barrier Reef, the world's largest coral reef system located
+            in Australia, is a vast and diverse ecosystem. However, it faces serious
+            threats from climate change, ocean acidification, and coral bleaching,
+            endangering its rich marine life.""",
+                """The Great Barrier Reef, a vast coral reef system off the coast of
+            Queensland, Australia, is the world's largest. It's a complex ecosystem
+            supporting diverse marine life, including endangered species. However,
+            climate change, ocean acidification, and coral bleaching are serious
+            threats to its survival.""",
+                """The Great Barrier Reef, the world's largest coral reef system off the
+            coast of Australia, is a vast and diverse ecosystem with thousands of
+            reefs and islands. It is home to a multitude of marine life, including
+            endangered species, but faces serious threats from climate change, ocean
+            acidification, and coral bleaching.""",
+            ],
+            "reference": [reference_summarization] * 3,
+        }
+    )
+    eval_task = EvalTask(
+        dataset=eval_dataset,
+        metrics=[
+            "rouge_1",
+            "rouge_2",
+            "rouge_l",
+            "rouge_l_sum",
+        ],
+    )
+    result = eval_task.evaluate()
+
+    print("Summary Metrics:\n")
+    for key, value in result.summary_metrics.items():
+        print(f"{key}: \t{value}")
+
+    print("\n\nMetrics Table:\n")
+    print(result.metrics_table)
+    # Example response:
+    #
+    # Summary Metrics:
+    #
+    # row_count:      3
+    # rouge_1/mean:   0.7191161666666667
+    # rouge_1/std:    0.06765143922270488
+    # rouge_2/mean:   0.5441118566666666
+    # ...
+    # Metrics Table:
+    #
+    #                                        response                         reference  ...  rouge_l/score  rouge_l_sum/score
+    # 0  The Great Barrier Reef, the world's ...  \n    The Great Barrier Reef, the ...  ...       0.577320           0.639175
+    # 1  The Great Barrier Reef, a vast coral...  \n    The Great Barrier Reef, the ...  ...       0.552381           0.666667
+    # 2  The Great Barrier Reef, the world's ...  \n    The Great Barrier Reef, the ...  ...       0.774775           0.774775
+    #
+    return result
+
+
+if __name__ == "__main__":
+    get_rouge_score()
diff --git a/genai/evaluation/noxfile_config.py b/genai/evaluation/noxfile_config.py
new file mode 100644
index 00000000000..1d2019a0b18
--- /dev/null
+++ b/genai/evaluation/noxfile_config.py
@@ -0,0 +1,42 @@
+# Copyright 2021 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Default TEST_CONFIG_OVERRIDE for python repos.
+
+# You can copy this file into your directory, then it will be imported from
+# the noxfile.py.
+
+# The source of truth:
+# https://github.com/GoogleCloudPlatform/python-docs-samples/blob/main/noxfile_config.py
+
+TEST_CONFIG_OVERRIDE = {
+    # You can opt out from the test for specific Python versions.
+    "ignored_versions": ["3.8", "3.9", "3.10", "3.12", "3.13"],
+    # Old samples are opted out of enforcing Python type hints
+    # All new samples should feature them
+    "enforce_type_hints": True,
+    # An envvar key for determining the project id to use. Change it
+    # to 'BUILD_SPECIFIC_GCLOUD_PROJECT' if you want to opt in using a
+    # build specific Cloud project. You can also use your own string
+    # to use your own Cloud project.
+    "gcloud_project_env": "GOOGLE_CLOUD_PROJECT",
+    # 'gcloud_project_env': 'BUILD_SPECIFIC_GCLOUD_PROJECT',
+    # If you need to use a specific version of pip,
+    # change pip_version_override to the string representation
+    # of the version number, for example, "20.2.4"
+    "pip_version_override": None,
+    # A dictionary you want to inject into your test. Don't put any
+    # secrets here. These values will override predefined values.
+    "envs": {},
+}
diff --git a/genai/evaluation/pairwise_summarization_quality.py b/genai/evaluation/pairwise_summarization_quality.py
new file mode 100644
index 00000000000..3d04ad5de46
--- /dev/null
+++ b/genai/evaluation/pairwise_summarization_quality.py
@@ -0,0 +1,119 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# [START aiplatform_genai_evaluation_pairwise_summarization_quality]
+
+import os
+
+import pandas as pd
+
+from google import genai
+from google.genai import types
+
+from vertexai.preview.evaluation import EvalResult
+from vertexai.evaluation import (
+    EvalTask,
+    PairwiseMetric,
+    MetricPromptTemplateExamples,
+)
+
+# TODO (developer) set GOOGLE_CLOUD_PROJECT and REGION_ID
+# environment variables before running.
+PROJECT_ID = os.getenv("GOOGLE_CLOUD_PROJECT")
+LOCATION = os.getenv("REGION_ID")
+BASELINE_MODEL = os.getenv("BASELINE_MODEL", "gemini-2.5-flash")
+CANDIDATE_MODEL = os.getenv("CANDIDATE_MODEL", "gemini-2.5-pro")
+
+PROMPT = """
+    Summarize the text such that a five-year-old can understand.
+
+    # Text
+
+    As part of a comprehensive initiative to tackle urban congestion and foster
+    sustainable urban living, a major city has revealed ambitious plans for an
+    extensive overhaul of its public transportation system. The project aims not
+    only to improve the efficiency and reliability of public transit but also to
+    reduce the city\'s carbon footprint and promote eco-friendly commuting options.
+    City officials anticipate that this strategic investment will enhance
+    accessibility for residents and visitors alike, ushering in a new era of
+    efficient, environmentally conscious urban transportation.
+    """
+
+def evaluate_output() -> EvalResult:
+    """
+    Evaluates a candidate model's summarization quality
+    against a baseline model using Vertex AI.
+    """
+
+    baseline_responses = []
+    candidate_responses = []
+
+    genai_client = genai.Client(vertexai=True, project=PROJECT_ID, location=LOCATION)
+
+    baseline_resp = genai_client.models.generate_content(
+        model="gemini-2.5-flash",
+        contents=PROMPT,
+        config=types.GenerateContentConfig(temperature=0.4)
+    )
+    baseline_responses.append(baseline_resp.text)
+
+    candidate_resp = genai_client.models.generate_content(
+        model="gemini-2.5-pro",
+        contents=PROMPT,
+        config=types.GenerateContentConfig(temperature=0.4)
+    )
+    candidate_responses.append(candidate_resp.text)
+
+    eval_df = pd.DataFrame({
+        "prompt": PROMPT,
+        "response": candidate_responses,
+        "baseline_model_response": baseline_responses
+    })
+
+    prompt_template = MetricPromptTemplateExamples.get_prompt_template(
+        "pairwise_summarization_quality"
+    )
+
+    pairwise_text_quality = PairwiseMetric(
+        metric="pairwise_summarization_quality",
+        metric_prompt_template=prompt_template,
+    )
+
+    eval_task = EvalTask(
+        dataset=eval_df,
+        metrics=[pairwise_text_quality],
+        experiment="pairwise-benchmark",
+    )
+
+    comparison_result = eval_task.evaluate()
+
+    pd.set_option('display.max_columns', None)
+    pd.set_option('display.max_colwidth', 250)
+
+    columns_to_print = [
+        "prompt", 
+        "baseline_model_response",
+        "response",
+        "pairwise_summarization_quality/pairwise_choice", 
+        "pairwise_summarization_quality/explanation"
+    ]
+    print(comparison_result.metrics_table[columns_to_print])
+
+    return comparison_result
+
+# [END aiplatform_genai_evaluation_pairwise_summarization_quality]
+
+
+if __name__ == "__main__":
+    evaluate_output()
diff --git a/genai/evaluation/requirements-test.txt b/genai/evaluation/requirements-test.txt
new file mode 100644
index 00000000000..b3b2b3ca4cc
--- /dev/null
+++ b/genai/evaluation/requirements-test.txt
@@ -0,0 +1 @@
+pytest==9.0.3
diff --git a/genai/evaluation/requirements.txt b/genai/evaluation/requirements.txt
new file mode 100644
index 00000000000..666fe9e6240
--- /dev/null
+++ b/genai/evaluation/requirements.txt
@@ -0,0 +1,4 @@
+pandas==2.3.3
+google-auth==2.55.2
+google-cloud-aiplatform==1.161.0
+google-genai==2.12.1
diff --git a/genai/evaluation/test_evaluation.py b/genai/evaluation/test_evaluation.py
new file mode 100644
index 00000000000..8263eb97709
--- /dev/null
+++ b/genai/evaluation/test_evaluation.py
@@ -0,0 +1,26 @@
+# Copyright 2024 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import get_rouge_score
+import pairwise_summarization_quality
+
+
+def test_create_evaluation_task() -> None:
+    response = get_rouge_score.get_rouge_score()
+    assert response
+
+
+def test_pairwise_evaluation_summarization_quality() -> None:
+    response = pairwise_summarization_quality.evaluate_output()
+    assert response
diff --git a/generative_ai/evaluation/get_rouge_score.py b/generative_ai/evaluation/get_rouge_score.py
index 323750f95d6..3e7206f95d5 100644
--- a/generative_ai/evaluation/get_rouge_score.py
+++ b/generative_ai/evaluation/get_rouge_score.py
@@ -11,21 +11,19 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
-# [START generativeaionvertexai_evaluation_get_rouge_score]
-
 import os
 
-import pandas as pd
-
-import vertexai
 from vertexai.preview.evaluation import EvalResult
-from vertexai.preview.evaluation import EvalTask
 
 PROJECT_ID = os.getenv("GOOGLE_CLOUD_PROJECT")
 
 
 def get_rouge_score() -> EvalResult:
+    # TODO To be refactored in b/536110099
+    import pandas as pd
+
+    import vertexai
+    from vertexai.preview.evaluation import EvalTask
 
     # TODO(developer): Update & uncomment line below
     # PROJECT_ID = "your-project-id"
@@ -93,7 +91,7 @@ def get_rouge_score() -> EvalResult:
     # 0  The Great Barrier Reef, the world's ...  \n    The Great Barrier Reef, the ...  ...       0.577320           0.639175
     # 1  The Great Barrier Reef, a vast coral...  \n    The Great Barrier Reef, the ...  ...       0.552381           0.666667
     # 2  The Great Barrier Reef, the world's ...  \n    The Great Barrier Reef, the ...  ...       0.774775           0.774775
-    # [END generativeaionvertexai_evaluation_get_rouge_score]
+
     return result
 
 
diff --git a/generative_ai/evaluation/noxfile_config.py b/generative_ai/evaluation/noxfile_config.py
index 1d2019a0b18..0973c8621c7 100644
--- a/generative_ai/evaluation/noxfile_config.py
+++ b/generative_ai/evaluation/noxfile_config.py
@@ -22,7 +22,7 @@
 
 TEST_CONFIG_OVERRIDE = {
     # You can opt out from the test for specific Python versions.
-    "ignored_versions": ["3.8", "3.9", "3.10", "3.12", "3.13"],
+    "ignored_versions": ["3.8", "3.9", "3.11", "3.12", "3.13"],
     # Old samples are opted out of enforcing Python type hints
     # All new samples should feature them
     "enforce_type_hints": True,
diff --git a/generative_ai/evaluation/pairwise_summarization_quality.py b/generative_ai/evaluation/pairwise_summarization_quality.py
index ea116a5cb8d..88c89871904 100644
--- a/generative_ai/evaluation/pairwise_summarization_quality.py
+++ b/generative_ai/evaluation/pairwise_summarization_quality.py
@@ -11,50 +11,28 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
-# [START generativeaionvertexai_evaluation_pairwise_summarization_quality]
-
 import os
 
-import pandas as pd
-
-from google import genai
-from google.genai import types
-
-import vertexai
 from vertexai.preview.evaluation import EvalResult
-from vertexai.evaluation import (
-    EvalTask,
-    PairwiseMetric,
-    MetricPromptTemplateExamples,
-)
-
-# TODO (developer) set GOOGLE_CLOUD_PROJECT and REGION_ID
-# environment variables before running.
-PROJECT_ID = os.getenv("GOOGLE_CLOUD_PROJECT")
-LOCATION = os.getenv("REGION_ID")
-BASELINE_MODEL = "gemini-2.5-flash"
-
 
-def custom_model_fn(prompt: str) -> str:
-    """Generates text from a prompt using the baseline Gemini model via Vertex AI."""
-
-    genai_client = genai.Client(vertexai=True, project=PROJECT_ID, location=LOCATION)
+PROJECT_ID = os.getenv("GOOGLE_CLOUD_PROJECT")
 
-    genai_config = types.GenerateContentConfig(temperature=0.4)
 
-    response = genai_client.models.generate_content(
-        model=BASELINE_MODEL, contents=prompt, config=genai_config
+def evaluate_output() -> EvalResult:
+    # [START generativeaionvertexai_evaluation_pairwise_summarization_quality]
+    import pandas as pd
+
+    import vertexai
+    from vertexai.generative_models import GenerativeModel
+    from vertexai.evaluation import (
+        EvalTask,
+        PairwiseMetric,
+        MetricPromptTemplateExamples,
     )
 
-    return response.text
-
-
-def evaluate_output() -> EvalResult:
-    """
-    Evaluates a candidate model's summarization quality
-    against a baseline model using Vertex AI.
-    """
+    # TODO(developer): Update & uncomment line below
+    # PROJECT_ID = "your-project-id"
+    vertexai.init(project=PROJECT_ID, location="us-central1")
 
     prompt = """
     Summarize the text such that a five-year-old can understand.
@@ -71,10 +49,16 @@ def evaluate_output() -> EvalResult:
     efficient, environmentally conscious urban transportation.
     """
 
-    vertexai.init(project=PROJECT_ID, location=LOCATION)
-
     eval_dataset = pd.DataFrame({"prompt": [prompt]})
 
+    # Baseline model for pairwise comparison
+    baseline_model = GenerativeModel("gemini-2.0-flash-lite-001")
+
+    # Candidate model for pairwise comparison
+    candidate_model = GenerativeModel(
+        "gemini-2.0-flash-001", generation_config={"temperature": 0.4}
+    )
+
     prompt_template = MetricPromptTemplateExamples.get_prompt_template(
         "pairwise_summarization_quality"
     )
@@ -82,7 +66,7 @@ def evaluate_output() -> EvalResult:
     summarization_quality_metric = PairwiseMetric(
         metric="pairwise_summarization_quality",
         metric_prompt_template=prompt_template,
-        baseline_model=BASELINE_MODEL,
+        baseline_model=baseline_model,
     )
 
     eval_task = EvalTask(
@@ -90,17 +74,13 @@ def evaluate_output() -> EvalResult:
         metrics=[summarization_quality_metric],
         experiment="pairwise-experiment",
     )
-    result = eval_task.evaluate(
-        model=custom_model_fn, experiment_run_name="genai-client-vs-legacy-baseline-6"
-    )
+    result = eval_task.evaluate(model=candidate_model)
 
     baseline_model_response = result.metrics_table["baseline_model_response"].iloc[0]
     candidate_model_response = result.metrics_table["response"].iloc[0]
-
     winner_model = result.metrics_table[
         "pairwise_summarization_quality/pairwise_choice"
     ].iloc[0]
-
     explanation = result.metrics_table[
         "pairwise_summarization_quality/explanation"
     ].iloc[0]
diff --git a/generative_ai/evaluation/requirements-test.txt b/generative_ai/evaluation/requirements-test.txt
index 598922c8c25..baa23bf9c3e 100644
--- a/generative_ai/evaluation/requirements-test.txt
+++ b/generative_ai/evaluation/requirements-test.txt
@@ -1,4 +1,4 @@
 backoff==2.2.1
-google-api-core==2.31.0
-pytest==9.0.3
-pytest-asyncio==1.4.0
+google-api-core==2.19.0
+pytest==9.0.3; python_version >= "3.10"
+pytest-asyncio==0.23.6
diff --git a/generative_ai/evaluation/requirements.txt b/generative_ai/evaluation/requirements.txt
index 842e7cc88b6..18ff8773d69 100644
--- a/generative_ai/evaluation/requirements.txt
+++ b/generative_ai/evaluation/requirements.txt
@@ -1,4 +1,13 @@
-pandas==2.3.3
-google-auth==2.55.2
-google-cloud-aiplatform[full]==1.161.0
-google-genai==2.12.1
+pandas==2.2.3; python_version == '3.7'
+pandas==2.2.3; python_version == '3.8'
+pandas==2.2.3; python_version > '3.8'
+pillow==12.2.0
+google-cloud-aiplatform[full]==1.157.0
+sentencepiece==0.2.1
+google-auth==2.38.0
+anthropic[vertex]==0.28.0
+langchain-core==1.4.0
+langchain-google-vertexai==3.2.3
+numpy<3
+openai==1.68.2
+immutabledict==4.2.0

From 1183bf99f477b6c27edf0a794cf75b43804f09a8 Mon Sep 17 00:00:00 2001
From: David del Real Sifuentes <deldavid@google.com>
Date: Mon, 20 Jul 2026 20:39:39 +0000
Subject: [PATCH 3/4] modified wrong file, undoing.

---
 generative_ai/evaluation/get_rouge_score.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/generative_ai/evaluation/get_rouge_score.py b/generative_ai/evaluation/get_rouge_score.py
index 3e7206f95d5..579c0931374 100644
--- a/generative_ai/evaluation/get_rouge_score.py
+++ b/generative_ai/evaluation/get_rouge_score.py
@@ -19,7 +19,7 @@
 
 
 def get_rouge_score() -> EvalResult:
-    # TODO To be refactored in b/536110099
+    # [START generativeaionvertexai_evaluation_get_rouge_score]
     import pandas as pd
 
     import vertexai
@@ -91,7 +91,7 @@ def get_rouge_score() -> EvalResult:
     # 0  The Great Barrier Reef, the world's ...  \n    The Great Barrier Reef, the ...  ...       0.577320           0.639175
     # 1  The Great Barrier Reef, a vast coral...  \n    The Great Barrier Reef, the ...  ...       0.552381           0.666667
     # 2  The Great Barrier Reef, the world's ...  \n    The Great Barrier Reef, the ...  ...       0.774775           0.774775
-
+    # [END generativeaionvertexai_evaluation_get_rouge_score]
     return result
 
 

From 00b951b8d478b847866c8b1587a113ab0feae11b Mon Sep 17 00:00:00 2001
From: David del Real Sifuentes <deldavid@google.com>
Date: Mon, 20 Jul 2026 22:15:37 +0000
Subject: [PATCH 4/4] linting issues fixed.

---
 .../pairwise_summarization_quality.py         | 38 ++++++++++---------
 genai/evaluation/requirements.txt             |  2 +-
 2 files changed, 22 insertions(+), 18 deletions(-)

diff --git a/genai/evaluation/pairwise_summarization_quality.py b/genai/evaluation/pairwise_summarization_quality.py
index 3d04ad5de46..260064da152 100644
--- a/genai/evaluation/pairwise_summarization_quality.py
+++ b/genai/evaluation/pairwise_summarization_quality.py
@@ -16,17 +16,17 @@
 
 import os
 
-import pandas as pd
-
 from google import genai
 from google.genai import types
 
-from vertexai.preview.evaluation import EvalResult
+import pandas as pd
+
 from vertexai.evaluation import (
     EvalTask,
-    PairwiseMetric,
     MetricPromptTemplateExamples,
+    PairwiseMetric,
 )
+from vertexai.preview.evaluation import EvalResult
 
 # TODO (developer) set GOOGLE_CLOUD_PROJECT and REGION_ID
 # environment variables before running.
@@ -50,6 +50,7 @@
     efficient, environmentally conscious urban transportation.
     """
 
+
 def evaluate_output() -> EvalResult:
     """
     Evaluates a candidate model's summarization quality
@@ -64,22 +65,24 @@ def evaluate_output() -> EvalResult:
     baseline_resp = genai_client.models.generate_content(
         model="gemini-2.5-flash",
         contents=PROMPT,
-        config=types.GenerateContentConfig(temperature=0.4)
+        config=types.GenerateContentConfig(temperature=0.4),
     )
     baseline_responses.append(baseline_resp.text)
 
     candidate_resp = genai_client.models.generate_content(
         model="gemini-2.5-pro",
         contents=PROMPT,
-        config=types.GenerateContentConfig(temperature=0.4)
+        config=types.GenerateContentConfig(temperature=0.4),
     )
     candidate_responses.append(candidate_resp.text)
 
-    eval_df = pd.DataFrame({
-        "prompt": PROMPT,
-        "response": candidate_responses,
-        "baseline_model_response": baseline_responses
-    })
+    eval_df = pd.DataFrame(
+        {
+            "prompt": PROMPT,
+            "response": candidate_responses,
+            "baseline_model_response": baseline_responses,
+        }
+    )
 
     prompt_template = MetricPromptTemplateExamples.get_prompt_template(
         "pairwise_summarization_quality"
@@ -93,25 +96,26 @@ def evaluate_output() -> EvalResult:
     eval_task = EvalTask(
         dataset=eval_df,
         metrics=[pairwise_text_quality],
-        experiment="pairwise-benchmark",
+        experiment="pairwise-benchmark2",
     )
 
     comparison_result = eval_task.evaluate()
 
-    pd.set_option('display.max_columns', None)
-    pd.set_option('display.max_colwidth', 250)
+    pd.set_option("display.max_columns", None)
+    pd.set_option("display.max_colwidth", 250)
 
     columns_to_print = [
-        "prompt", 
+        "prompt",
         "baseline_model_response",
         "response",
-        "pairwise_summarization_quality/pairwise_choice", 
-        "pairwise_summarization_quality/explanation"
+        "pairwise_summarization_quality/pairwise_choice",
+        "pairwise_summarization_quality/explanation",
     ]
     print(comparison_result.metrics_table[columns_to_print])
 
     return comparison_result
 
+
 # [END aiplatform_genai_evaluation_pairwise_summarization_quality]
 
 
diff --git a/genai/evaluation/requirements.txt b/genai/evaluation/requirements.txt
index 666fe9e6240..f039d20dab4 100644
--- a/genai/evaluation/requirements.txt
+++ b/genai/evaluation/requirements.txt
@@ -1,4 +1,4 @@
 pandas==2.3.3
 google-auth==2.55.2
-google-cloud-aiplatform==1.161.0
+google-cloud-aiplatform[evaluation]==1.161.0
 google-genai==2.12.1