diff --git a/.github/workflows/tests.yml b/.github/workflows/tests.yml
new file mode 100644
index 0000000..c1300f6
--- /dev/null
+++ b/.github/workflows/tests.yml
@@ -0,0 +1,31 @@
+name: Run Tests
+
+on:
+  workflow_dispatch:
+  pull_request:
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    environment: "Run Tests"
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.13"
+          cache: pip
+
+      - name: Install dependencies
+        run: pip install -r requirements.txt && pip install -e .
+
+      - name: Run tests
+        run: pytest
+        env:
+          YOU_API_KEY: ${{ secrets.YOU_API_KEY }}
+          EXA_API_KEY: ${{ secrets.EXA_API_KEY }}
+          SERP_API_KEY: ${{ secrets.SERP_API_KEY }}
+          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
+          PARALLEL_API_KEY: ${{ secrets.PARALLEL_API_KEY }}
+          TAVILY_API_KEY: ${{ secrets.TAVILY_API_KEY }}
+          GOOGLE_GEMINI_API_KEY: ${{ secrets.GOOGLE_GEMINI_API_KEY }}
diff --git a/tests/test_simpleqa.py b/tests/test_simpleqa.py
index 8d9eae7..ee2eb34 100644
--- a/tests/test_simpleqa.py
+++ b/tests/test_simpleqa.py
@@ -55,7 +55,7 @@ async def test_simpleqa_runner(test_results_cleanup):
     4. Metrics are calculated correctly
     """
     # Create test arguments
-    num_problems = 10
+    num_problems = 2
     results_dir = get_test_results_dir()
     args = argparse.Namespace(
         samplers=[sampler.sampler_name for sampler in samplers.SAMPLERS],
@@ -63,7 +63,6 @@ async def test_simpleqa_runner(test_results_cleanup):
         limit=num_problems,  # Test with small subset for speed
         batch_size=10,
         max_concurrent_tasks=10,
-        num_results=5,
         clean=True,
     )
 
@@ -84,7 +83,7 @@ async def test_simpleqa_runner(test_results_cleanup):
         expected_columns = [
             "query",
             "internal_response_time_ms",
-            "end_to_end_time_ms",
+            "request_response_time_ms",
             "evaluation_result",
         ]
         for col in expected_columns:
@@ -114,8 +113,8 @@ async def test_simpleqa_runner(test_results_cleanup):
             == args.samplers.sort()
         )
         assert "accuracy_score" in df_metrics.columns
-        assert "avg_internal_latency" in df_metrics.columns
-        assert "avg_end_to_end_latency" in df_metrics.columns
+        assert "p50_internal_latency" in df_metrics.columns
+        assert "p50_request_response_latency" in df_metrics.columns
         assert "problem_count" in df_metrics.columns
 
 
@@ -127,7 +126,7 @@ async def test_simpleqa_runner_resume_capability(test_results_cleanup):
     This test verifies that if a run is interrupted, it can continue
     from where it left off without re-processing completed queries.
     """
-    num_problems = 10
+    num_problems = 2
     results_dir = get_test_results_dir()
     # Create test arguments for first run (partial)
     args = argparse.Namespace(
@@ -153,7 +152,7 @@ async def test_simpleqa_runner_resume_capability(test_results_cleanup):
         ), f"Expected {num_problems} results from first run, got {first_run_count}"
 
     # Second run with more queries (should add new results)
-    args.limit = 5
+    args.limit = 2
     args.clean = False  # Don't clean, resume from existing
     await run_evals(args, results_dir=results_dir)
     for sampler in args.samplers: