diff --git a/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh b/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
index 3212b660ec356..a394046d2c8fe 100755
--- a/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
@@ -150,7 +150,7 @@ run_and_track_test 9 "test_multimodal.py" \
 run_and_track_test 10 "test_pallas.py" \
     "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_pallas.py"
 run_and_track_test 11 "test_struct_output_generate.py" \
-    "python3 -m pytest -s -v /workspace/vllm/tests/v1/entrypoints/llm/test_struct_output_generate.py"
+    "python3 -m pytest -s -v /workspace/vllm/tests/v1/entrypoints/llm/test_struct_output_generate.py -k 'not test_structured_output_with_reasoning_matrices'"
 run_and_track_test 12 "test_moe_pallas.py" \
     "python3 -m pytest -s -v /workspace/vllm/tests/tpu/test_moe_pallas.py"
 run_and_track_test 13 "test_lora.py" \
diff --git a/tests/v1/tpu/test_spmd_model_weight_loading.py b/tests/v1/tpu/test_spmd_model_weight_loading.py
index d36edfc3fb618..916325e41b922 100644
--- a/tests/v1/tpu/test_spmd_model_weight_loading.py
+++ b/tests/v1/tpu/test_spmd_model_weight_loading.py
@@ -45,11 +45,14 @@ def _get_spmd_mesh():
     return MESH
 
 
-@pytest.mark.parametrize("model", [
-    "Qwen/Qwen2-1.5B-Instruct",
-    "meta-llama/Llama-3.1-8B-Instruct",
-    "meta-llama/Llama-3.1-70B-Instruct",
-])
+@pytest.mark.parametrize(
+    "model",
+    [
+        "Qwen/Qwen2-1.5B-Instruct",
+        # Skip large models due to CI runner disk space limitations
+        # "meta-llama/Llama-3.1-8B-Instruct",
+        # "meta-llama/Llama-3.1-70B-Instruct",
+    ])
 def test_tpu_model_loader(model):
     # Skip the 70B test if there are less than 8 chips
     # TODO: Query using torch xla API, the query API is not working