diff --git a/benchmarks/multi_node/amd_utils/models.yaml b/benchmarks/multi_node/amd_utils/models.yaml
index 2bbdd91d6..ebc185025 100644
--- a/benchmarks/multi_node/amd_utils/models.yaml
+++ b/benchmarks/multi_node/amd_utils/models.yaml
@@ -39,7 +39,7 @@
 
 DeepSeek-V3:
   base_flags: "--decode-log-interval 1000 --log-level warning --watchdog-timeout 3600 --ep-dispatch-algorithm fake --load-balance-method round_robin --kv-cache-dtype fp8_e4m3 --attention-backend aiter --disaggregation-transfer-backend mori"
-  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1"
+  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1 --speculative-num-draft-tokens 3"
   dp_flags: "--moe-a2a-backend mori --deepep-mode normal --enable-dp-attention --moe-dense-tp-size 1 --enable-dp-lm-head"
   prefill:
     mem_fraction_static: 0.8
@@ -70,7 +70,7 @@ DeepSeek-V3:
 
 DeepSeek-V3-0324:
   base_flags: "--decode-log-interval 1000 --log-level warning --watchdog-timeout 3600 --ep-dispatch-algorithm fake --load-balance-method round_robin --kv-cache-dtype fp8_e4m3 --attention-backend aiter --disaggregation-transfer-backend mori"
-  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1"
+  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1 --speculative-num-draft-tokens 3"
   dp_flags: "--moe-a2a-backend mori --deepep-mode normal --enable-dp-attention --moe-dense-tp-size 1 --enable-dp-lm-head"
   prefill:
     mem_fraction_static: 0.8
@@ -101,7 +101,7 @@ DeepSeek-V3-0324:
 
 DeepSeek-R1:
   base_flags: "--decode-log-interval 1000 --log-level warning --watchdog-timeout 3600 --ep-dispatch-algorithm fake --load-balance-method round_robin --kv-cache-dtype fp8_e4m3 --attention-backend aiter --disaggregation-transfer-backend mori"
-  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1"
+  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1 --speculative-num-draft-tokens 3"
   dp_flags: "--moe-a2a-backend mori --deepep-mode normal --enable-dp-attention --moe-dense-tp-size 1 --enable-dp-lm-head"
   prefill:
     mem_fraction_static: 0.8
@@ -132,7 +132,7 @@ DeepSeek-R1:
 
 DeepSeek-R1-0528:
   base_flags: "--decode-log-interval 1000 --log-level warning --watchdog-timeout 3600 --ep-dispatch-algorithm fake --load-balance-method round_robin --kv-cache-dtype fp8_e4m3 --attention-backend aiter --disaggregation-transfer-backend mori"
-  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1"
+  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1 --speculative-num-draft-tokens 3"
   dp_flags: "--moe-a2a-backend mori --deepep-mode normal --enable-dp-attention --moe-dense-tp-size 1 --enable-dp-lm-head"
   prefill:
     mem_fraction_static: 0.8
@@ -163,7 +163,7 @@ DeepSeek-R1-0528:
 
 DeepSeek-R1-0528-MXFP4-Preview:
   base_flags: "--decode-log-interval 1000 --log-level warning --watchdog-timeout 3600 --ep-dispatch-algorithm fake --load-balance-method round_robin --kv-cache-dtype fp8_e4m3 --attention-backend aiter --disaggregation-transfer-backend mori"
-  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1"
+  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1 --speculative-num-draft-tokens 3"
   dp_flags: "--moe-a2a-backend mori --deepep-mode normal --enable-dp-attention --moe-dense-tp-size 1 --enable-dp-lm-head"
   prefill:
     mem_fraction_static: 0.8
@@ -194,7 +194,7 @@ DeepSeek-R1-0528-MXFP4-Preview:
 
 DeepSeek-R1-0528-MXFP4:
   base_flags: "--decode-log-interval 1000 --log-level warning --watchdog-timeout 3600 --ep-dispatch-algorithm fake --load-balance-method round_robin --kv-cache-dtype fp8_e4m3 --attention-backend aiter --disaggregation-transfer-backend mori"
-  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1"
+  mtp_flags: "--speculative-algorithm NEXTN --speculative-eagle-topk 1 --speculative-num-draft-tokens 3"
   dp_flags: "--moe-a2a-backend mori --deepep-mode normal --enable-dp-attention --moe-dense-tp-size 1 --enable-dp-lm-head"
   prefill:
     mem_fraction_static: 0.8
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 2090a1b25..0986a5c6d 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -961,3 +961,10 @@
   description:
     - "Extend MiniMax M2.5 FP8 single-node config for H200 with vLLM v0.16.0 (TP8)"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/869
+
+- config-keys:
+    - dsr1-fp4-mi355x-sglang-disagg-mtp
+    - dsr1-fp8-mi355x-sglang-disagg-mtp
+  description:
+    - "test"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXX