diff --git a/test/srt/ascend/test_ascend_compile_graph_tp1_bf16.py b/test/srt/ascend/test_ascend_compile_graph_tp1_bf16.py
index deaf4a5e0..4c6a1e8d0 100644
--- a/test/srt/ascend/test_ascend_compile_graph_tp1_bf16.py
+++ b/test/srt/ascend/test_ascend_compile_graph_tp1_bf16.py
@@ -15,7 +15,7 @@ from sglang.test.test_utils import (
 )
 
 TEST_MODEL_MATRIX = {
-    "Qwen/Qwen2.5-7B-Instruct": {
+    "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct": {
         "accuracy": 0.84,
         "latency": 150,
         "output_throughput": 30,
diff --git a/test/srt/ascend/test_ascend_graph_tp1_bf16.py b/test/srt/ascend/test_ascend_graph_tp1_bf16.py
index 95c6b7bcf..e181f7aac 100644
--- a/test/srt/ascend/test_ascend_graph_tp1_bf16.py
+++ b/test/srt/ascend/test_ascend_graph_tp1_bf16.py
@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
 )
 
 TEST_MODEL_MATRIX = {
-    "Qwen/Qwen2.5-7B-Instruct": {
+    "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct": {
         "accuracy": 0.85,
         "latency": 150,
         "output_throughput": 30,
diff --git a/test/srt/ascend/test_ascend_graph_tp2_bf16.py b/test/srt/ascend/test_ascend_graph_tp2_bf16.py
index f7c3c6537..da65f5c6a 100644
--- a/test/srt/ascend/test_ascend_graph_tp2_bf16.py
+++ b/test/srt/ascend/test_ascend_graph_tp2_bf16.py
@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
 )
 
 TEST_MODEL_MATRIX = {
-    "Qwen/Qwen2.5-7B-Instruct": {
+    "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct": {
         "accuracy": 0.85,
         "latency": 180,
         "output_throughput": 20,
diff --git a/test/srt/ascend/test_ascend_hicache_mha.py b/test/srt/ascend/test_ascend_hicache_mha.py
index d6f1aa9c2..2a91af17c 100644
--- a/test/srt/ascend/test_ascend_hicache_mha.py
+++ b/test/srt/ascend/test_ascend_hicache_mha.py
@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
 )
 
 TEST_MODEL_MATRIX = {
-    "Qwen/Qwen2.5-7B-Instruct": {
+    "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct": {
         "accuracy": 0.85,
         "latency": 150,
         "output_throughput": 30,
diff --git a/test/srt/ascend/test_ascend_piecewise_graph_prefill.py b/test/srt/ascend/test_ascend_piecewise_graph_prefill.py
index 7f13d13e8..44122b716 100644
--- a/test/srt/ascend/test_ascend_piecewise_graph_prefill.py
+++ b/test/srt/ascend/test_ascend_piecewise_graph_prefill.py
@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
     run_bench_one_batch,
 )
 
-MODEL = "Qwen/Qwen2.5-7B-Instruct"
+MODEL = "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct"
 GSM8K_EXP_ACCURACY = 0.84
 EXP_PREFILL_LATENCY = 0.045
 TOKENS_TO_CAPTURE = [i for i in range(128, 4096, 128)]
diff --git a/test/srt/ascend/test_ascend_sampling_backend.py b/test/srt/ascend/test_ascend_sampling_backend.py
index 7b6307912..f0eee21a3 100644
--- a/test/srt/ascend/test_ascend_sampling_backend.py
+++ b/test/srt/ascend/test_ascend_sampling_backend.py
@@ -16,7 +16,7 @@ from sglang.test.test_utils import (
 class TestAscendSamplingBackend(CustomTestCase):
     @classmethod
     def setUpClass(cls):
-        cls.model = "Qwen/Qwen2.5-7B-Instruct"
+        cls.model = "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct"
         cls.base_url = DEFAULT_URL_FOR_TEST
         cls.process = popen_launch_server(
             cls.model,
diff --git a/test/srt/ascend/test_ascend_tp1_bf16.py b/test/srt/ascend/test_ascend_tp1_bf16.py
index fd0f96e73..abc960953 100644
--- a/test/srt/ascend/test_ascend_tp1_bf16.py
+++ b/test/srt/ascend/test_ascend_tp1_bf16.py
@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
 )
 
 TEST_MODEL_MATRIX = {
-    "Qwen/Qwen2.5-7B-Instruct": {
+    "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct": {
         "accuracy": 0.84,
         "latency": 150,
         "output_throughput": 30,
diff --git a/test/srt/ascend/test_ascend_tp2_bf16.py b/test/srt/ascend/test_ascend_tp2_bf16.py
index d5e141c9f..64f52cd4d 100644
--- a/test/srt/ascend/test_ascend_tp2_bf16.py
+++ b/test/srt/ascend/test_ascend_tp2_bf16.py
@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
 )
 
 TEST_MODEL_MATRIX = {
-    "Qwen/Qwen2.5-7B-Instruct": {
+    "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct": {
         "accuracy": 0.85,
         "latency": 180,
         "output_throughput": 20,
diff --git a/test/srt/ascend/test_ascend_tp2_fia_bf16.py b/test/srt/ascend/test_ascend_tp2_fia_bf16.py
index 6e2f0d8d0..5f82bb47f 100644
--- a/test/srt/ascend/test_ascend_tp2_fia_bf16.py
+++ b/test/srt/ascend/test_ascend_tp2_fia_bf16.py
@@ -13,7 +13,7 @@ from sglang.test.test_utils import (
 )
 
 TEST_MODEL_MATRIX = {
-    "Qwen/Qwen2.5-7B-Instruct": {
+    "/root/.cache/modelscope/hub/models/Qwen/Qwen2.5-7B-Instruct": {
         "accuracy": 0.85,
         "latency": 180,
         "output_throughput": 20,