diff --git a/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4-MTP-Align.yaml b/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4-MTP-Align.yaml
new file mode 100644
index 000000000000..4373f1bbfa14
--- /dev/null
+++ b/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4-MTP-Align.yaml
@@ -0,0 +1,9 @@
+model_name: "Qwen/Qwen3-Next-80B-A3B-Instruct-FP8"
+accuracy_threshold: 0.85
+num_questions: 1319
+num_fewshot: 5
+server_args: >-
+  --max-model-len 4096
+  --tensor-parallel-size 4
+  --speculative-config '{"method":"qwen3_next_mtp","num_speculative_tokens":2}'
+  --enable-prefix-caching
diff --git a/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4-MTP.yaml b/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4-MTP.yaml
new file mode 100644
index 000000000000..8736b929e5ad
--- /dev/null
+++ b/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4-MTP.yaml
@@ -0,0 +1,8 @@
+model_name: "Qwen/Qwen3-Next-80B-A3B-Instruct-FP8"
+accuracy_threshold: 0.85
+num_questions: 1319
+num_fewshot: 5
+server_args: >-
+  --max-model-len 4096
+  --tensor-parallel-size 4
+  --speculative-config '{"method":"qwen3_next_mtp","num_speculative_tokens":2}'
diff --git a/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4.yaml b/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4.yaml
new file mode 100644
index 000000000000..c9dfc64131d0
--- /dev/null
+++ b/tests/evals/gsm8k/configs/hybrid/Qwen3-Next-FP8-TP4.yaml
@@ -0,0 +1,7 @@
+model_name: "Qwen/Qwen3-Next-80B-A3B-Instruct-FP8"
+accuracy_threshold: 0.85
+num_questions: 1319
+num_fewshot: 5
+server_args: >-
+  --max-model-len 4096
+  --tensor-parallel-size 4
diff --git a/tests/evals/gsm8k/configs/hybrid/models-h100.txt b/tests/evals/gsm8k/configs/hybrid/models-h100.txt
new file mode 100644
index 000000000000..a84e6b5fc105
--- /dev/null
+++ b/tests/evals/gsm8k/configs/hybrid/models-h100.txt
@@ -0,0 +1,3 @@
+Qwen3-Next-FP8-TP4.yaml
+Qwen3-Next-FP8-TP4-MTP.yaml
+Qwen3-Next-FP8-TP4-MTP-Align.yaml