diff --git a/test/srt/test_gpt_oss_common.py b/python/sglang/test/gpt_oss_common.py
similarity index 100%
rename from test/srt/test_gpt_oss_common.py
rename to python/sglang/test/gpt_oss_common.py
diff --git a/test/srt/test_deterministic.py b/test/registered/core/test_deterministic.py
similarity index 92%
rename from test/srt/test_deterministic.py
rename to test/registered/core/test_deterministic.py
index 4d6c8e608..3bbe9f965 100644
--- a/test/srt/test_deterministic.py
+++ b/test/registered/core/test_deterministic.py
@@ -9,11 +9,14 @@ test into unit tests so that's easily reproducible in CI.
 
 import unittest
 
+from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_deterministic_utils import (
     COMMON_SERVER_ARGS,
     TestDeterministicBase,
 )
 
+register_cuda_ci(est_time=228, suite="stage-b-test-small-1-gpu")
+
 
 class TestFlashinferDeterministic(TestDeterministicBase):
     # Test with flashinfer attention backend
diff --git a/test/srt/test_gpt_oss_1gpu.py b/test/registered/core/test_gpt_oss_1gpu.py
similarity index 73%
rename from test/srt/test_gpt_oss_1gpu.py
rename to test/registered/core/test_gpt_oss_1gpu.py
index d3fc06931..503257652 100644
--- a/test/srt/test_gpt_oss_1gpu.py
+++ b/test/registered/core/test_gpt_oss_1gpu.py
@@ -1,6 +1,10 @@
 import unittest
 
-from test_gpt_oss_common import BaseTestGptOss
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
+from sglang.test.gpt_oss_common import BaseTestGptOss
+
+register_cuda_ci(est_time=402, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=750, suite="stage-b-test-small-1-gpu-amd")
 
 
 class TestGptOss1Gpu(BaseTestGptOss):
diff --git a/test/srt/test_hidden_states.py b/test/registered/core/test_hidden_states.py
similarity index 97%
rename from test/srt/test_hidden_states.py
rename to test/registered/core/test_hidden_states.py
index 2046ce529..0933df815 100644
--- a/test/srt/test_hidden_states.py
+++ b/test/registered/core/test_hidden_states.py
@@ -4,8 +4,11 @@ import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
 import sglang as sgl
+from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import DEFAULT_SMALL_MODEL_NAME_FOR_TEST, CustomTestCase
 
+register_cuda_ci(est_time=55, suite="stage-b-test-small-1-gpu")
+
 
 class TestHiddenState(CustomTestCase):
     def test_return_hidden_states(self):
diff --git a/test/srt/test_input_embeddings.py b/test/registered/core/test_input_embeddings.py
similarity index 96%
rename from test/srt/test_input_embeddings.py
rename to test/registered/core/test_input_embeddings.py
index 7efd66565..bf80328f8 100644
--- a/test/srt/test_input_embeddings.py
+++ b/test/registered/core/test_input_embeddings.py
@@ -7,6 +7,7 @@ import requests
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
 from sglang.srt.utils import kill_process_tree
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
@@ -15,6 +16,9 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
+register_cuda_ci(est_time=38, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=38, suite="stage-b-test-small-1-gpu-amd")
+
 
 class TestInputEmbeds(CustomTestCase):
     @classmethod
diff --git a/test/srt/test_io_struct.py b/test/registered/core/test_io_struct.py
similarity index 99%
rename from test/srt/test_io_struct.py
rename to test/registered/core/test_io_struct.py
index a8efebb23..ac7b8c990 100644
--- a/test/srt/test_io_struct.py
+++ b/test/registered/core/test_io_struct.py
@@ -2,12 +2,16 @@ import copy
 import unittest
 
 from sglang.srt.managers.io_struct import GenerateReqInput
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
     DEFAULT_URL_FOR_TEST,
     CustomTestCase,
 )
 
+register_cuda_ci(est_time=8, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=8, suite="stage-b-test-small-1-gpu-amd")
+
 
 class TestGenerateReqInputNormalization(CustomTestCase):
     """Test the normalization of GenerateReqInput for batch processing and different input formats."""
diff --git a/test/srt/test_model_hooks.py b/test/registered/core/test_model_hooks.py
similarity index 95%
rename from test/srt/test_model_hooks.py
rename to test/registered/core/test_model_hooks.py
index 473aafb3e..ae14a2abd 100644
--- a/test/srt/test_model_hooks.py
+++ b/test/registered/core/test_model_hooks.py
@@ -6,8 +6,12 @@ import torch.nn as nn
 
 from sglang.srt.model_executor.hook_manager import register_forward_hooks
 from sglang.srt.server_args import ServerArgs
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
+register_cuda_ci(est_time=6, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=10, suite="stage-b-test-small-1-gpu-amd")
+
 HOOK_CALLS = []
 
 
diff --git a/test/srt/test_page_size.py b/test/registered/core/test_page_size.py
similarity index 85%
rename from test/srt/test_page_size.py
rename to test/registered/core/test_page_size.py
index f56e1ed6b..b28bd3c7d 100644
--- a/test/srt/test_page_size.py
+++ b/test/registered/core/test_page_size.py
@@ -3,6 +3,7 @@ import unittest
 from types import SimpleNamespace
 
 from sglang.srt.utils import kill_process_tree
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.run_eval import run_eval
 from sglang.test.test_utils import (
     DEFAULT_MODEL_NAME_FOR_TEST,
@@ -12,6 +13,9 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
+register_cuda_ci(est_time=60, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=60, suite="stage-b-test-small-1-gpu-amd")
+
 
 class TestPageSize(CustomTestCase):
     @classmethod
diff --git a/test/srt/test_request_queue_validation.py b/test/registered/core/test_request_queue_validation.py
similarity index 93%
rename from test/srt/test_request_queue_validation.py
rename to test/registered/core/test_request_queue_validation.py
index 3dd41b0d4..c76e006ca 100644
--- a/test/srt/test_request_queue_validation.py
+++ b/test/registered/core/test_request_queue_validation.py
@@ -4,6 +4,7 @@ import re
 import unittest
 
 from sglang.srt.utils import kill_process_tree
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
@@ -16,6 +17,9 @@ from sglang.test.test_utils import (
     send_generate_requests,
 )
 
+register_cuda_ci(est_time=47, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=70, suite="stage-b-test-small-1-gpu-amd")
+
 
 class TestMaxQueuedRequests(CustomTestCase):
     @classmethod
diff --git a/test/srt/test_score_api.py b/test/registered/core/test_score_api.py
similarity index 99%
rename from test/srt/test_score_api.py
rename to test/registered/core/test_score_api.py
index 757af86de..6bf3c034c 100644
--- a/test/srt/test_score_api.py
+++ b/test/registered/core/test_score_api.py
@@ -4,8 +4,11 @@ import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
 from sglang.srt.entrypoints.engine import Engine
+from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import DEFAULT_SMALL_MODEL_NAME_FOR_TEST, CustomTestCase
 
+register_cuda_ci(est_time=260, suite="stage-b-test-small-1-gpu")
+
 TEST_MODEL_NAME = DEFAULT_SMALL_MODEL_NAME_FOR_TEST
 
 
diff --git a/test/srt/test_server_args.py b/test/registered/core/test_server_args.py
similarity index 98%
rename from test/srt/test_server_args.py
rename to test/registered/core/test_server_args.py
index 04c6ce22e..91e90a3e4 100644
--- a/test/srt/test_server_args.py
+++ b/test/registered/core/test_server_args.py
@@ -3,8 +3,12 @@ import unittest
 from unittest.mock import patch
 
 from sglang.srt.server_args import PortArgs, ServerArgs, prepare_server_args
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
+register_cuda_ci(est_time=9, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=1, suite="stage-b-test-small-1-gpu-amd")
+
 
 class TestPrepareServerArgs(CustomTestCase):
     def test_prepare_server_args(self):
diff --git a/test/srt/test_srt_endpoint.py b/test/registered/core/test_srt_endpoint.py
similarity index 99%
rename from test/srt/test_srt_endpoint.py
rename to test/registered/core/test_srt_endpoint.py
index 59a8c3c46..92aeff84c 100644
--- a/test/srt/test_srt_endpoint.py
+++ b/test/registered/core/test_srt_endpoint.py
@@ -17,6 +17,7 @@ import requests
 
 from sglang.srt.sampling.custom_logit_processor import CustomLogitProcessor
 from sglang.srt.utils import kill_process_tree
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
@@ -26,6 +27,9 @@ from sglang.test.test_utils import (
     run_logprob_check,
 )
 
+register_cuda_ci(est_time=127, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=130, suite="stage-b-test-small-1-gpu-amd")
+
 
 class TestSRTEndpoint(CustomTestCase):
     @classmethod
diff --git a/test/srt/test_srt_engine.py b/test/registered/core/test_srt_engine.py
similarity index 97%
rename from test/srt/test_srt_engine.py
rename to test/registered/core/test_srt_engine.py
index d370f6290..e4227669a 100644
--- a/test/srt/test_srt_engine.py
+++ b/test/registered/core/test_srt_engine.py
@@ -14,6 +14,7 @@ import sglang as sgl
 from sglang.bench_offline_throughput import BenchArgs, throughput_test
 from sglang.srt.server_args import ServerArgs
 from sglang.srt.utils.hf_transformers_utils import get_tokenizer
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.few_shot_gsm8k_engine import run_eval
 from sglang.test.test_utils import (
     DEFAULT_SMALL_EMBEDDING_MODEL_NAME_FOR_TEST,
@@ -21,6 +22,9 @@ from sglang.test.test_utils import (
     CustomTestCase,
 )
 
+register_cuda_ci(est_time=252, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=261, suite="stage-b-test-small-1-gpu-amd")
+
 
 class TestSRTEngine(CustomTestCase):
 
diff --git a/test/srt/run_suite.py b/test/srt/run_suite.py
index e091abb96..d8d8fb70a 100644
--- a/test/srt/run_suite.py
+++ b/test/srt/run_suite.py
@@ -9,24 +9,12 @@ from sglang.test.ci.ci_utils import TestFile, run_unittest_files
 # NOTE: please sort the test cases alphabetically by the test file name
 suites = {
     "per-commit-1-gpu": [
-        TestFile("test_deterministic.py", 228),
         TestFile("test_evs.py", 20),
         TestFile("test_external_models.py", 30),
-        TestFile("test_gpt_oss_1gpu.py", 402),
-        TestFile("test_hidden_states.py", 55),
-        TestFile("test_input_embeddings.py", 38),
-        TestFile("test_io_struct.py", 8),
         TestFile("test_jinja_template_utils.py", 7),
-        TestFile("test_model_hooks.py", 6),
         TestFile("test_modelopt_loader.py", 11),
         TestFile("test_multi_tokenizer.py", 230),
-        TestFile("test_page_size.py", 60),
-        TestFile("test_request_queue_validation.py", 47),
-        TestFile("test_score_api.py", 260),
-        TestFile("test_server_args.py", 9),
         TestFile("test_skip_tokenizer_init.py", 77),
-        TestFile("test_srt_endpoint.py", 127),
-        TestFile("test_srt_engine.py", 252),
         TestFile("test_utils_update_weights.py", 29),
         TestFile("test_video_utils.py", 5),
         TestFile("test_modelopt_export.py", 9),
@@ -98,7 +86,6 @@ suites = {
         TestFile("test_bench_one_batch.py"),
         TestFile("test_bench_serving.py"),
         TestFile("test_eval_accuracy_large.py"),
-        TestFile("test_gpt_oss_common.py"),
         TestFile("test_moe_eval_accuracy_large.py"),
         TestFile("test_profile_v2.py"),
         TestFile("models/test_ministral3_models.py"),
@@ -120,18 +107,10 @@ suite_amd = {
         # TestFile("lora/test_lora_qwen3.py", 97), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/13107
         TestFile("test_bench_typebaseddispatcher.py", 10),
         TestFile("test_external_models.py", 45),
-        TestFile("test_input_embeddings.py", 38),
-        TestFile("test_io_struct.py", 8),
         TestFile("test_jinja_template_utils.py", 1),
-        TestFile("test_model_hooks.py", 10),
         TestFile("test_multi_tokenizer.py", 345),
-        TestFile("test_page_size.py", 60),
-        TestFile("test_request_queue_validation.py", 70),
         TestFile("test_rope_rocm.py", 3),
-        TestFile("test_server_args.py", 1),
         TestFile("test_skip_tokenizer_init.py", 117),
-        TestFile("test_srt_endpoint.py", 130),
-        TestFile("test_srt_engine.py", 261),
         # TestFile("test_torch_compile_moe.py", 210), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/13107
         TestFile("test_type_based_dispatcher.py", 10),
         TestFile("test_video_utils.py", 8),
@@ -143,9 +122,7 @@ suite_amd = {
         # TestFile("test_vision_chunked_prefill.py", 175), # Disabled temporarily and track in #7701
         # TestFile("test_wave_attention_backend.py", 150), # Disabled temporarily, see https://github.com/sgl-project/sglang/issues/11127
     ],
-    "per-commit-amd-mi35x": [
-        TestFile("test_gpt_oss_1gpu.py", 750),
-    ],
+    "per-commit-amd-mi35x": [],
     "per-commit-2-gpu-amd": [
         TestFile("test_data_parallelism.py", 73),
         TestFile("test_load_weights_from_remote_instance.py", 72),
diff --git a/test/srt/test_gpt_oss_4gpu.py b/test/srt/test_gpt_oss_4gpu.py
index da787c6fb..f2e212994 100644
--- a/test/srt/test_gpt_oss_4gpu.py
+++ b/test/srt/test_gpt_oss_4gpu.py
@@ -1,6 +1,6 @@
 import unittest
 
-from test_gpt_oss_common import BaseTestGptOss
+from sglang.test.gpt_oss_common import BaseTestGptOss
 
 
 class TestGptOss4Gpu(BaseTestGptOss):