diff --git a/.github/workflows/tests.yml b/.github/workflows/tests.yml
index 0f4c1ae31..a1b864305 100644
--- a/.github/workflows/tests.yml
+++ b/.github/workflows/tests.yml
@@ -22,6 +22,8 @@ jobs:
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
+        with:
+          fetch-depth: 0  # Fetch all history for proper diff
 
       - name: Restore global uv cache
         id: cache-restore
@@ -49,22 +51,29 @@ jobs:
         uses: actions/cache/restore@v4
         with:
           path: .test_durations_py*
-          key: test-durations-py${{ matrix.python-version }}-
-          restore-keys: |
-            test-durations-py${{ matrix.python-version }}-
+          key: test-durations-py${{ matrix.python-version }}
 
       - name: Run tests (group ${{ matrix.group }} of 8)
         run: |
           PYTHON_VERSION_SAFE=$(echo "${{ matrix.python-version }}" | tr '.' '_')
           DURATION_FILE=".test_durations_py${PYTHON_VERSION_SAFE}"
           
-          if [ -f "$DURATION_FILE" ]; then
-            echo "Using cached test durations for optimal splitting"
-            DURATIONS_ARG="--durations-path=${DURATION_FILE}"
-          else
-            echo "No cached durations found, tests will be split evenly"
-            DURATIONS_ARG=""
-          fi
+          # Temporarily always skip cached durations to fix test splitting
+          # When durations don't match, pytest-split runs duplicate tests instead of splitting
+          echo "Using even test splitting (duration cache disabled until fix merged)"
+          DURATIONS_ARG=""
+          
+          # Original logic (disabled temporarily):
+          # if [ ! -f "$DURATION_FILE" ]; then
+          #   echo "No cached durations found, tests will be split evenly"
+          #   DURATIONS_ARG=""
+          # elif git diff origin/${{ github.base_ref }}...HEAD --name-only 2>/dev/null | grep -q "^tests/.*\.py$"; then
+          #   echo "Test files have changed, skipping cached durations to avoid mismatches"
+          #   DURATIONS_ARG=""
+          # else
+          #   echo "No test changes detected, using cached test durations for optimal splitting"
+          #   DURATIONS_ARG="--durations-path=${DURATION_FILE}"
+          # fi
           
           uv run pytest \
             --block-network \
diff --git a/.github/workflows/update-test-durations.yml b/.github/workflows/update-test-durations.yml
index daa1decfd..13f1ecd69 100644
--- a/.github/workflows/update-test-durations.yml
+++ b/.github/workflows/update-test-durations.yml
@@ -58,7 +58,7 @@ jobs:
         uses: actions/cache/save@v4
         with:
           path: .test_durations_py*
-          key: test-durations-py${{ matrix.python-version }}-${{ github.sha }}
+          key: test-durations-py${{ matrix.python-version }}
 
       - name: Save uv caches
         if: steps.cache-restore.outputs.cache-hit != 'true'