diff --git a/tests/integration/defs/accuracy/references/mmlu.yaml b/tests/integration/defs/accuracy/references/mmlu.yaml
index 122913c74c6..bf70fa6d24c 100644
--- a/tests/integration/defs/accuracy/references/mmlu.yaml
+++ b/tests/integration/defs/accuracy/references/mmlu.yaml
@@ -203,6 +203,7 @@ Qwen3/Qwen3-8B:
   - spec_dec_algo: Eagle
     accuracy: 76.12
 Qwen3/Qwen3-30B-A3B:
+  - accuracy: 79.53
   - quant_algo: FP8_BLOCK_SCALES
     accuracy: 79.53
   - quant_algo: FP8
diff --git a/tests/integration/defs/accuracy/test_disaggregated_serving.py b/tests/integration/defs/accuracy/test_disaggregated_serving.py
index 86595362095..88b97ae84d5 100644
--- a/tests/integration/defs/accuracy/test_disaggregated_serving.py
+++ b/tests/integration/defs/accuracy/test_disaggregated_serving.py
@@ -915,4 +915,6 @@ def test_mixed_ctx_gen_model(self, ctx_pp, gen_tp):
                                  gen_tp=gen_tp,
                                  test_sets=[GSM8K, MMLU],
                                  ctx_model=ctx_model,
-                                 gen_model=gen_model)
+                                 gen_model=gen_model,
+                                 ctx_instances=1,
+                                 gen_instances=1)