diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p1d-dep4-dep8-c308-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p1d-dep4-dep8-c308-stp.yaml
new file mode 100644
index 0000000000..1cb6ab0de4
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p1d-dep4-dep8-c308-stp.yaml
@@ -0,0 +1,128 @@
+name: disagg-B200-1p1d-dep4-dep8-c308-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 1
+  decode_workers: 1
+  gpus_per_decode: 8
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        - 8
+        - 16
+        - 24
+        - 32
+        enable_padding: true
+      enable_attention_dp: true
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.92
+      max_batch_size: 32
+      max_num_tokens: 32
+      max_seq_len: 9256
+      moe_config:
+        backend: CUTEDSL
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 8
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 8
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 308
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p4d-dep4-tep8-c24-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p4d-dep4-tep8-c24-stp.yaml
new file mode 100644
index 0000000000..49a71471de
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p4d-dep4-tep8-c24-stp.yaml
@@ -0,0 +1,124 @@
+name: disagg-B200-1p4d-dep4-tep8-c24-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 4
+  decode_workers: 4
+  gpus_per_decode: 8
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        enable_padding: true
+      enable_attention_dp: false
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 4
+      max_num_tokens: 4
+      max_seq_len: 9256
+      moe_config:
+        backend: TRTLLM
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 8
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 8
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 24
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p4d-dep4-tep8-c4-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p4d-dep4-tep8-c4-stp.yaml
new file mode 100644
index 0000000000..cabec31b18
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p4d-dep4-tep8-c4-stp.yaml
@@ -0,0 +1,124 @@
+name: disagg-B200-1p4d-dep4-tep8-c4-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 4
+  decode_workers: 4
+  gpus_per_decode: 8
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        enable_padding: true
+      enable_attention_dp: false
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 1
+      max_num_tokens: 1
+      max_seq_len: 9256
+      moe_config:
+        backend: TRTLLM
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 8
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 8
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 4
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c115-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c115-stp.yaml
new file mode 100644
index 0000000000..298279a010
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c115-stp.yaml
@@ -0,0 +1,126 @@
+name: disagg-B200-1p5d-dep4-tep4-c115-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 3
+  decode_workers: 5
+  gpus_per_decode: 4
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        - 8
+        - 16
+        enable_padding: true
+      enable_attention_dp: false
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 16
+      max_num_tokens: 16
+      max_seq_len: 9256
+      moe_config:
+        backend: TRTLLM
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 4
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 4
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 115
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c195-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c195-stp.yaml
new file mode 100644
index 0000000000..f31a7da80e
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c195-stp.yaml
@@ -0,0 +1,128 @@
+name: disagg-B200-1p5d-dep4-tep4-c195-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 3
+  decode_workers: 5
+  gpus_per_decode: 4
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        - 8
+        - 16
+        - 24
+        - 32
+        enable_padding: true
+      enable_attention_dp: false
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 32
+      max_num_tokens: 32
+      max_seq_len: 9256
+      moe_config:
+        backend: TRTLLM
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 4
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 4
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 195
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c30-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c30-stp.yaml
new file mode 100644
index 0000000000..4c4c833192
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c30-stp.yaml
@@ -0,0 +1,124 @@
+name: disagg-B200-1p5d-dep4-tep4-c30-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 3
+  decode_workers: 5
+  gpus_per_decode: 4
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        enable_padding: true
+      enable_attention_dp: false
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 4
+      max_num_tokens: 4
+      max_seq_len: 9256
+      moe_config:
+        backend: TRTLLM
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 4
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 4
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 30
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c5-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c5-stp.yaml
new file mode 100644
index 0000000000..619c1bb709
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c5-stp.yaml
@@ -0,0 +1,124 @@
+name: disagg-B200-1p5d-dep4-tep4-c5-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 3
+  decode_workers: 5
+  gpus_per_decode: 4
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        enable_padding: true
+      enable_attention_dp: false
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 1
+      max_num_tokens: 1
+      max_seq_len: 9256
+      moe_config:
+        backend: TRTLLM
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 4
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 4
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 5
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c60-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c60-stp.yaml
new file mode 100644
index 0000000000..1a8b4f2988
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-1p5d-dep4-tep4-c60-stp.yaml
@@ -0,0 +1,125 @@
+name: disagg-B200-1p5d-dep4-tep4-c60-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 1
+  gpus_per_prefill: 4
+  decode_nodes: 3
+  decode_workers: 5
+  gpus_per_decode: 4
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        - 8
+        enable_padding: true
+      enable_attention_dp: false
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 8
+      max_num_tokens: 8
+      max_seq_len: 9256
+      moe_config:
+        backend: TRTLLM
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 4
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 4
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 60
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-2p1d-dep4-dep8-c615-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-2p1d-dep4-dep8-c615-stp.yaml
new file mode 100644
index 0000000000..abcce625e7
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-2p1d-dep4-dep8-c615-stp.yaml
@@ -0,0 +1,132 @@
+name: disagg-B200-2p1d-dep4-dep8-c615-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 1
+  prefill_workers: 2
+  gpus_per_prefill: 4
+  decode_nodes: 1
+  decode_workers: 1
+  gpus_per_decode: 8
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        - 8
+        - 16
+        - 24
+        - 32
+        - 40
+        - 48
+        - 56
+        - 64
+        enable_padding: true
+      enable_attention_dp: true
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.92
+      max_batch_size: 64
+      max_num_tokens: 64
+      max_seq_len: 9256
+      moe_config:
+        backend: CUTEDSL
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 8
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 8
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 615
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-3p1d-dep4-dep8-c1127-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-3p1d-dep4-dep8-c1127-stp.yaml
new file mode 100644
index 0000000000..316a88f23a
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-3p1d-dep4-dep8-c1127-stp.yaml
@@ -0,0 +1,140 @@
+name: disagg-B200-3p1d-dep4-dep8-c1127-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 2
+  prefill_workers: 3
+  gpus_per_prefill: 4
+  decode_nodes: 1
+  decode_workers: 1
+  gpus_per_decode: 8
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        - 8
+        - 16
+        - 24
+        - 32
+        - 40
+        - 48
+        - 56
+        - 64
+        - 72
+        - 80
+        - 88
+        - 96
+        - 104
+        - 112
+        - 120
+        - 128
+        enable_padding: true
+      enable_attention_dp: true
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.92
+      max_batch_size: 128
+      max_num_tokens: 128
+      max_seq_len: 9256
+      moe_config:
+        backend: CUTEDSL
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 8
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 8
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 1127
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-4p1d-dep4-dep8-c2151-stp.yaml b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-4p1d-dep4-dep8-c2151-stp.yaml
new file mode 100644
index 0000000000..893a725cae
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4/disagg-B200-4p1d-dep4-dep8-c2151-stp.yaml
@@ -0,0 +1,156 @@
+name: disagg-B200-4p1d-dep4-dep8-c2151-stp
+model:
+  path: kimik2.5-fp4
+  container: dynamo-trtllm
+  precision: fp4
+identity:
+  model:
+    repo: nvidia/Kimi-K2.5-NVFP4
+  container:
+    image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+slurm:
+  time_limit: 04:00:00
+dynamo:
+  install: true
+  hash: 4025cb6a8992d19fd19eead36065e08ea5301f35
+  request_plane: tcp
+health_check:
+  max_attempts: 540
+  interval_seconds: 10
+resources:
+  gpu_type: b200
+  gpus_per_node: 8
+  prefill_nodes: 2
+  prefill_workers: 4
+  gpus_per_prefill: 4
+  decode_nodes: 1
+  decode_workers: 1
+  gpus_per_decode: 8
+backend:
+  type: trtllm
+  prefill_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  decode_environment:
+    TLLM_LOG_LEVEL: INFO
+    TRTLLM_SERVER_DISABLE_GC: '1'
+    TRTLLM_WORKER_DISABLE_GC: '1'
+    TRTLLM_ENABLE_PDL: '1'
+    NCCL_GRAPH_MIXING_SUPPORT: '0'
+    MIMALLOC_PURGE_DELAY: '0'
+    UCX_TLS: rc,cuda_ipc,cuda_copy,sm,self,tcp
+  trtllm_config:
+    prefill:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config: null
+      disable_overlap_scheduler: true
+      enable_attention_dp: true
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.9
+      max_batch_size: 2
+      max_num_tokens: 16386
+      max_seq_len: 8232
+      moe_config:
+        backend: CUTEDSL
+      moe_expert_parallel_size: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      tensor_parallel_size: 4
+      trust_remote_code: true
+    decode:
+      cache_transceiver_config:
+        backend: DEFAULT
+        kv_transfer_timeout_ms: 600000
+        max_tokens_in_buffer: 68736
+      cuda_graph_config:
+        batch_sizes:
+        - 1
+        - 2
+        - 4
+        - 8
+        - 16
+        - 24
+        - 32
+        - 40
+        - 48
+        - 56
+        - 64
+        - 72
+        - 80
+        - 88
+        - 96
+        - 104
+        - 112
+        - 120
+        - 128
+        - 136
+        - 144
+        - 152
+        - 160
+        - 168
+        - 176
+        - 184
+        - 192
+        - 200
+        - 208
+        - 216
+        - 224
+        - 232
+        - 240
+        - 248
+        - 256
+        enable_padding: true
+      enable_attention_dp: true
+      enable_lm_head_tp_in_adp: false
+      kv_cache_config:
+        dtype: fp8
+        enable_block_reuse: false
+        free_gpu_memory_fraction: 0.92
+      max_batch_size: 256
+      max_num_tokens: 256
+      max_seq_len: 9256
+      moe_config:
+        backend: CUTEDSL
+        use_low_precision_moe_combine: true
+      moe_expert_parallel_size: 8
+      num_postprocess_workers: 4
+      nvfp4_gemm_config:
+        allowed_backends:
+        - cutlass
+        - cublaslt
+        - cutedsl
+        - cuda_core
+      pipeline_parallel_size: 1
+      print_iter_log: true
+      stream_interval: 100
+      tensor_parallel_size: 8
+      trust_remote_code: true
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  env:
+    ETCD_LEASE_TTL: '120'
+benchmark:
+  type: sa-bench
+  isl: 8192
+  osl: 1024
+  concurrencies:
+  - 2151
+  req_rate: inf
+  num_prompts_mult: 20
diff --git a/configs/nvidia-master.yaml b/configs/nvidia-master.yaml
index cc31fc0e47..a167f49270 100644
--- a/configs/nvidia-master.yaml
+++ b/configs/nvidia-master.yaml
@@ -4266,6 +4266,167 @@ dsr1-fp4-b200-dynamo-sglang-mtp:
           ep: 8
           dp-attn: true
 
+kimik2.5-fp4-b200-dynamo-trt:
+  image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21
+  model: nvidia/Kimi-K2.5-NVFP4
+  model-prefix: kimik2.5
+  runner: b200-multinode
+  precision: fp4
+  framework: dynamo-trt
+  router: { name: dynamo-router, version: "v1.3.0rc21" }
+  kv-p2p-transfer: nixl
+  multinode: true
+  disagg: true
+  scenarios:
+    fixed-seq-len:
+    - isl: 8192
+      osl: 1024
+      search-space:
+      # Non-MTP configurations (STP)
+      - conc-list: [ 4 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p4d-dep4-tep8-c4-stp.yaml"
+        decode:
+          num-worker: 4
+          tp: 8
+          ep: 8
+          dp-attn: false
+      - conc-list: [ 24 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p4d-dep4-tep8-c24-stp.yaml"
+        decode:
+          num-worker: 4
+          tp: 8
+          ep: 8
+          dp-attn: false
+      - conc-list: [ 5 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p5d-dep4-tep4-c5-stp.yaml"
+        decode:
+          num-worker: 5
+          tp: 4
+          ep: 4
+          dp-attn: false
+      - conc-list: [ 30 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p5d-dep4-tep4-c30-stp.yaml"
+        decode:
+          num-worker: 5
+          tp: 4
+          ep: 4
+          dp-attn: false
+      - conc-list: [ 60 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p5d-dep4-tep4-c60-stp.yaml"
+        decode:
+          num-worker: 5
+          tp: 4
+          ep: 4
+          dp-attn: false
+      - conc-list: [ 115 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p5d-dep4-tep4-c115-stp.yaml"
+        decode:
+          num-worker: 5
+          tp: 4
+          ep: 4
+          dp-attn: false
+      - conc-list: [ 195 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p5d-dep4-tep4-c195-stp.yaml"
+        decode:
+          num-worker: 5
+          tp: 4
+          ep: 4
+          dp-attn: false
+      - conc-list: [ 308 ]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-1p1d-dep4-dep8-c308-stp.yaml"
+        decode:
+          num-worker: 1
+          tp: 8
+          ep: 8
+          dp-attn: true
+      - conc-list: [ 615 ]
+        prefill:
+          num-worker: 2
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-2p1d-dep4-dep8-c615-stp.yaml"
+        decode:
+          num-worker: 1
+          tp: 8
+          ep: 8
+          dp-attn: true
+      - conc-list: [ 1127 ]
+        prefill:
+          num-worker: 3
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-3p1d-dep4-dep8-c1127-stp.yaml"
+        decode:
+          num-worker: 1
+          tp: 8
+          ep: 8
+          dp-attn: true
+      - conc-list: [ 2151 ]
+        prefill:
+          num-worker: 4
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/trtllm/kimi-k25-nvfp4/b200-fp4/disagg-B200-4p1d-dep4-dep8-c2151-stp.yaml"
+        decode:
+          num-worker: 1
+          tp: 8
+          ep: 8
+          dp-attn: true
+
 kimik2.5-fp4-gb200-dynamo-trt:
   image: nvcr.io#nvidia/ai-dynamo/tensorrtllm-runtime:1.3.0-dev.1-cuda13
   model: nvidia/Kimi-K2.5-NVFP4
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 1822ffcb70..e82277b6f0 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -5019,3 +5019,12 @@
     - "dsv4-fp4-b200-vllm-agentic-mtp: MTP twins (--speculative-config {\"method\":\"mtp\",\"num_speculative_tokens\":3}) of the aggregate arms -- TP8 GPU-resident, TP8 SimpleCPU, DEP8 SimpleCPU, DEP8 Mooncake, each mirroring its non-MTP conc-list; routed via spec-decoding=mtp to dsv4_fp4_b200_vllm_mtp.sh."
     - "MTP capture sizes are in TOKENS: --max-cudagraph-capture-size scaled to MAX_NUM_SEQS*(1+num_speculative_tokens) (= MAX_NUM_SEQS*4) so FULL_DECODE_ONLY covers the largest decode batches."
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2259
+
+- config-keys:
+    - kimik2.5-fp4-b200-dynamo-trt
+  description:
+    - "Add Kimi K2.5 NVFP4 B200 disaggregated TRT-LLM benchmarks via Dynamo (11 STP configs, ISL 8k OSL 1k)"
+    - "Container: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc21 with Dynamo installed from source (hash 4025cb6a8992d19fd19eead36065e08ea5301f35)"
+    - "Topologies: 1p4d-dep4-tep8 (conc 4/24), 1p5d-dep4-tep4 (conc 5/30/60/115/195), 1p1d-dep4-dep8 (conc 308), 2p1d-dep4-dep8 (conc 615), 3p1d-dep4-dep8 (conc 1127), 4p1d-dep4-dep8 (conc 2151)"
+    - "Runner updated to clone srt-slurm at sa-submission-q2-2026 and copy local recipes into recipes/trtllm/kimi-k25-nvfp4/b200-fp4/"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2249
diff --git a/runners/launch_b200-dgxc.sh b/runners/launch_b200-dgxc.sh
index 58051e6405..a276644575 100644
--- a/runners/launch_b200-dgxc.sh
+++ b/runners/launch_b200-dgxc.sh
@@ -128,6 +128,12 @@ if [[ "$IS_MULTINODE" == "true" ]]; then
         git checkout a98738de9b2233459b5456e9ed71af09ce893f92
         mkdir -p recipes/sglang/dsr1/b200-fp4
         cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/sglang/dsr1/b200-fp4" recipes/sglang/dsr1/b200-fp4
+    elif [[ $FRAMEWORK == "dynamo-trt" && $MODEL_PREFIX == "kimik2.5" && $PRECISION == "fp4" ]]; then
+        git clone https://github.com/NVIDIA/srt-slurm.git "$SRT_REPO_DIR"
+        cd "$SRT_REPO_DIR" || exit 1
+        git checkout v1.0.29
+        mkdir -p recipes/trtllm/kimi-k25-nvfp4/b200-fp4
+        cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/trtllm/kimi-k2.5/disagg/trtllm_dynamo/b200-fp4" recipes/trtllm/kimi-k25-nvfp4/b200-fp4
     else
         git clone https://github.com/NVIDIA/srt-slurm.git "$SRT_REPO_DIR"
         cd "$SRT_REPO_DIR" || exit 1