diff --git a/tests/integration/defs/perf/test_perf_sanity.py b/tests/integration/defs/perf/test_perf_sanity.py
index b6c7737cbe5a..4a24d6f748d3 100644
--- a/tests/integration/defs/perf/test_perf_sanity.py
+++ b/tests/integration/defs/perf/test_perf_sanity.py
@@ -1391,7 +1391,9 @@ def _parse_disagg_config_file(self, config_file_path: str, config_file: str):
         for concurrency in concurrency_values:
             client_config_data = {
                 "concurrency": concurrency,
-                "iterations": benchmark.get("multi_round", 1),
+                "iterations": 1
+                if benchmark_mode == "gen_only"
+                else benchmark.get("multi_round", 1),
                 "isl": benchmark.get("input_length", 1024),
                 "osl": osl,
                 "random_range_ratio": benchmark.get("benchmark_ratio", 0.0),
diff --git a/tests/integration/test_lists/waives.txt b/tests/integration/test_lists/waives.txt
index 7ced2db0c3f8..ff2a3a8429f1 100644
--- a/tests/integration/test_lists/waives.txt
+++ b/tests/integration/test_lists/waives.txt
@@ -325,9 +325,7 @@ disaggregated/test_disaggregated.py::test_disaggregated_gpt_oss_120b_harmony[gpt
 accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_bfloat16_4gpus[pp4-mtp_nextn=0-attention_dp=False-cuda_graph=False-overlap_scheduler=False-torch_compile=False] SKIP (https://nvbugs/6050489)
 disaggregated/test_disaggregated.py::test_disaggregated_overlap_gen_first[ctx_pp1-TinyLlama-1.1B-Chat-v1.0] SKIP (https://nvbugs/6057459)
 disaggregated/test_disaggregated.py::test_disaggregated_overlap_gen_first[ctx_pp4-TinyLlama-1.1B-Chat-v1.0] SKIP (https://nvbugs/6057460)
-perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-v32-fp4_32k4k_con1_ctx1_dep4_gen1_tep8_eplb0_mtp3_ccb-UCX] SKIP (https://nvbugs/5844149)
-perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-v32-fp4_32k4k_con2048_ctx1_dep4_gen1_dep32_eplb288_mtp1_ccb-UCX] SKIP (https://nvbugs/5844149)
-perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-v32-fp4_32k4k_con256_ctx1_dep4_gen1_dep32_eplb0_mtp3_ccb-UCX] SKIP (https://nvbugs/5844149)
+perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-v32-fp4_32k4k_con2048_ctx1_dep4_gen1_dep32_eplb288_mtp1_ccb-UCX] SKIP (https://nvbugs/6085022)
 perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-r1-fp4_128k8k_con128_ctx1_pp8_gen1_dep16_eplb0_mtp2_ccb-UCX] SKIP (https://nvbugs/6060119)
 perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-r1-fp4_128k8k_con64_ctx1_pp8_gen1_dep32_eplb0_mtp3_ccb-UCX] SKIP (https://nvbugs/6060119)
 full:sm89/accuracy/test_llm_api_pytorch.py::TestGemma3_1BInstruct::test_fp8_prequantized[torch_compile=False] SKIP (https://nvbugs/6070878)
diff --git a/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con1_ctx1_dep4_gen1_tep8_eplb0_mtp3_ccb-UCX.yaml b/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con1_ctx1_dep4_gen1_tep8_eplb0_mtp3_ccb-UCX.yaml
index 861b1fab55d1..faaf881af078 100644
--- a/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con1_ctx1_dep4_gen1_tep8_eplb0_mtp3_ccb-UCX.yaml
+++ b/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con1_ctx1_dep4_gen1_tep8_eplb0_mtp3_ccb-UCX.yaml
@@ -48,24 +48,27 @@ worker_config:
   gen:
     print_iter_log: true
     max_batch_size: 1
-    max_num_tokens: 256
+    max_num_tokens: 8
+    max_input_len: 32784
+    max_seq_len: 40960
     tensor_parallel_size: 8
     moe_expert_parallel_size: 8
     pipeline_parallel_size: 1
     context_parallel_size: 1
     enable_attention_dp: false
+    enable_lm_head_tp_in_adp: false
     cuda_graph_config:
       enable_padding: true
       max_batch_size: 1
     kv_cache_config:
       enable_block_reuse: false
-      free_gpu_memory_fraction: 0.9
+      free_gpu_memory_fraction: 0.25
       dtype: fp8
       tokens_per_block: 64
     moe_config:
       backend: TRTLLM
     cache_transceiver_config:
-      max_tokens_in_buffer: 32768
+      max_tokens_in_buffer: 32832
       backend: UCX
     disable_overlap_scheduler: true
     speculative_config: &id001
@@ -84,6 +87,8 @@ worker_config:
     print_iter_log: true
     max_batch_size: 1
     max_num_tokens: 32784
+    max_input_len: 32784
+    max_seq_len: 32832
     tensor_parallel_size: 4
     moe_expert_parallel_size: 4
     pipeline_parallel_size: 1
@@ -93,14 +98,14 @@ worker_config:
     cuda_graph_config: null
     kv_cache_config:
       enable_block_reuse: false
-      free_gpu_memory_fraction: 0.6
+      free_gpu_memory_fraction: 0.4
       dtype: fp8
       tokens_per_block: 64
     moe_config:
       backend: CUTEDSL
       use_low_precision_moe_combine: true
     cache_transceiver_config:
-      max_tokens_in_buffer: 32768
+      max_tokens_in_buffer: 32832
       backend: UCX
     disable_overlap_scheduler: true
     speculative_config: *id001
diff --git a/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con2048_ctx1_dep4_gen1_dep32_eplb288_mtp1_ccb-UCX.yaml b/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con2048_ctx1_dep4_gen1_dep32_eplb288_mtp1_ccb-UCX.yaml
index 502f14dc896d..a2b8f7b1a8f4 100644
--- a/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con2048_ctx1_dep4_gen1_dep32_eplb288_mtp1_ccb-UCX.yaml
+++ b/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con2048_ctx1_dep4_gen1_dep32_eplb288_mtp1_ccb-UCX.yaml
@@ -49,6 +49,8 @@ worker_config:
     print_iter_log: true
     max_batch_size: 64
     max_num_tokens: 128
+    max_input_len: 32784
+    max_seq_len: 40960
     tensor_parallel_size: 32
     moe_expert_parallel_size: 32
     pipeline_parallel_size: 1
@@ -60,7 +62,7 @@ worker_config:
       max_batch_size: 64
     kv_cache_config:
       enable_block_reuse: false
-      free_gpu_memory_fraction: 0.85
+      free_gpu_memory_fraction: 0.5
       dtype: fp8
       tokens_per_block: 64
     moe_config:
@@ -70,7 +72,7 @@ worker_config:
         num_slots: 288
         layer_updates_per_iter: 1
     cache_transceiver_config:
-      max_tokens_in_buffer: 32768
+      max_tokens_in_buffer: 32832
       backend: UCX
     disable_overlap_scheduler: true
     speculative_config: &id001
@@ -88,6 +90,8 @@ worker_config:
     print_iter_log: true
     max_batch_size: 1
     max_num_tokens: 32784
+    max_input_len: 32784
+    max_seq_len: 32832
     tensor_parallel_size: 4
     moe_expert_parallel_size: 4
     pipeline_parallel_size: 1
@@ -97,14 +101,14 @@ worker_config:
     cuda_graph_config: null
     kv_cache_config:
       enable_block_reuse: false
-      free_gpu_memory_fraction: 0.6
+      free_gpu_memory_fraction: 0.4
       dtype: fp8
       tokens_per_block: 64
     moe_config:
       backend: CUTEDSL
       use_low_precision_moe_combine: true
     cache_transceiver_config:
-      max_tokens_in_buffer: 32768
+      max_tokens_in_buffer: 32832
       backend: UCX
     disable_overlap_scheduler: true
     speculative_config: *id001
diff --git a/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con256_ctx1_dep4_gen1_dep32_eplb0_mtp3_ccb-UCX.yaml b/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con256_ctx1_dep4_gen1_dep32_eplb0_mtp3_ccb-UCX.yaml
index 24eb58b3b976..bdb9d1ded5ef 100644
--- a/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con256_ctx1_dep4_gen1_dep32_eplb0_mtp3_ccb-UCX.yaml
+++ b/tests/scripts/perf-sanity/disaggregated/gb200_deepseek-v32-fp4_32k4k_con256_ctx1_dep4_gen1_dep32_eplb0_mtp3_ccb-UCX.yaml
@@ -48,7 +48,9 @@ worker_config:
   gen:
     print_iter_log: true
     max_batch_size: 8
-    max_num_tokens: 256
+    max_num_tokens: 64
+    max_input_len: 32784
+    max_seq_len: 40960
     tensor_parallel_size: 32
     moe_expert_parallel_size: 32
     pipeline_parallel_size: 1
@@ -60,14 +62,14 @@ worker_config:
       max_batch_size: 8
     kv_cache_config:
       enable_block_reuse: false
-      free_gpu_memory_fraction: 0.85
+      free_gpu_memory_fraction: 0.5
       dtype: fp8
       tokens_per_block: 64
     moe_config:
       backend: CUTEDSL
       use_low_precision_moe_combine: true
     cache_transceiver_config:
-      max_tokens_in_buffer: 32768
+      max_tokens_in_buffer: 32832
       backend: UCX
     disable_overlap_scheduler: true
     speculative_config: &id001
@@ -85,6 +87,8 @@ worker_config:
     print_iter_log: true
     max_batch_size: 1
     max_num_tokens: 32784
+    max_input_len: 32784
+    max_seq_len: 32832
     tensor_parallel_size: 4
     moe_expert_parallel_size: 4
     pipeline_parallel_size: 1
@@ -94,14 +98,14 @@ worker_config:
     cuda_graph_config: null
     kv_cache_config:
       enable_block_reuse: false
-      free_gpu_memory_fraction: 0.6
+      free_gpu_memory_fraction: 0.4
       dtype: fp8
       tokens_per_block: 64
     moe_config:
       backend: CUTEDSL
       use_low_precision_moe_combine: true
     cache_transceiver_config:
-      max_tokens_in_buffer: 32768
+      max_tokens_in_buffer: 32832
       backend: UCX
     disable_overlap_scheduler: true
     speculative_config: *id001