[CI] Enable weight prefetch for 8-gpu-h200 basic tests (#25684)

test/registered/8-gpu-models/test_minimax_m25_basic.py

@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
     write_github_step_summary,
 )
 
-register_cuda_ci(est_time=290, stage="base-c", runner_config="8-gpu-h200")
+register_cuda_ci(est_time=250, stage="base-c", runner_config="8-gpu-h200")
 
 MINIMAX_M25_MODEL_PATH = "MiniMaxAI/MiniMax-M2.5"
 
@@ -36,6 +36,7 @@ class TestMiniMaxM25Basic(CustomTestCase):
             "minimax-append-think",
             "--model-loader-extra-config",
             '{"enable_multithread_load": true, "num_threads": 64}',
+            "--weight-loader-prefetch-checkpoints",
         ]
         cls.process = popen_launch_server(
             cls.model,

test/registered/radix_cache/test_unified_radix_cache_kl_hicache.py

@@ -24,7 +24,7 @@ MAMBA_TRACK_INTERVAL = 128
 DSV4_FLASH_MODEL = "sgl-project/DeepSeek-V4-Flash-FP8"
 DSV4_FLASH_LAUNCH_TIMEOUT = 3600
 
-register_cuda_ci(est_time=768, stage="base-c", runner_config="8-gpu-h200")
+register_cuda_ci(est_time=745, stage="base-c", runner_config="8-gpu-h200")
 
 
 class TestUnifiedMambaHiCache(UnifiedRadixTreeTestMixin, CustomTestCase):
@@ -72,6 +72,7 @@ class TestUnifiedMambaHiCache(UnifiedRadixTreeTestMixin, CustomTestCase):
                 "500",
                 "--max-running-requests",
                 "4",
+                "--weight-loader-prefetch-checkpoints",
             ],
             env={"SGLANG_ENABLE_UNIFIED_RADIX_TREE": "1"},
         )