diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-agentic.yaml
new file mode 100644
index 0000000000..421a096254
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-1p1d-tep4-tp4-c1-fp4-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: false
+  prefill_nodes: 1
+  decode_nodes: 1
+  prefill_workers: 1
+  decode_workers: 1
+  gpus_per_prefill: 4
+  gpus_per_decode: 4
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: true
+      all2all-backend: "flashinfer_nvlink_one_sided"
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-eval-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-eval-agentic.yaml
new file mode 100644
index 0000000000..3bed52725e
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-eval-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-1p1d-tep4-tp4-c1-fp4-eval-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: false
+  prefill_nodes: 1
+  decode_nodes: 1
+  prefill_workers: 1
+  decode_workers: 1
+  gpus_per_prefill: 4
+  gpus_per_decode: 4
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: true
+      all2all-backend: "flashinfer_nvlink_one_sided"
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-agentic.yaml
new file mode 100644
index 0000000000..fb4a6bf759
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-1p1d-tp2-tp4-c20-c24-fp4-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: false
+  prefill_nodes: 1
+  decode_nodes: 1
+  prefill_workers: 1
+  decode_workers: 1
+  gpus_per_prefill: 2
+  gpus_per_decode: 4
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 1
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-eval-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-eval-agentic.yaml
new file mode 100644
index 0000000000..755d2a8a12
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-eval-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-1p1d-tp2-tp4-c20-c24-fp4-eval-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: false
+  prefill_nodes: 1
+  decode_nodes: 1
+  prefill_workers: 1
+  decode_workers: 1
+  gpus_per_prefill: 2
+  gpus_per_decode: 4
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 1
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-agentic.yaml
new file mode 100644
index 0000000000..7bf7fbb74e
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-agentic.yaml
@@ -0,0 +1,166 @@
+name: "minimax-m3-vllm-disagg-gb300-1p3d-dep4-tp4-c24-fp4-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: false
+  prefill_nodes: 1
+  decode_nodes: 3
+  prefill_workers: 1
+  decode_workers: 3
+  gpus_per_prefill: 4
+  gpus_per_decode: 4
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 1
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 1
+      data-parallel-size: 4
+      data-parallel-rpc-port: 13345
+      enable-expert-parallel: true
+      all2all-backend: "flashinfer_nvlink_one_sided"
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-eval-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-eval-agentic.yaml
new file mode 100644
index 0000000000..261f1fe6d2
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-eval-agentic.yaml
@@ -0,0 +1,166 @@
+name: "minimax-m3-vllm-disagg-gb300-1p3d-dep4-tp4-c24-fp4-eval-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: false
+  prefill_nodes: 1
+  decode_nodes: 3
+  prefill_workers: 1
+  decode_workers: 3
+  gpus_per_prefill: 4
+  gpus_per_decode: 4
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 1
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 1
+      data-parallel-size: 4
+      data-parallel-rpc-port: 13345
+      enable-expert-parallel: true
+      all2all-backend: "flashinfer_nvlink_one_sided"
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 4
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-agentic.yaml
new file mode 100644
index 0000000000..8193a1f919
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-1p3d-tp2-tp2-c48-fp4-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: true
+  prefill_nodes: 1
+  decode_nodes: 3
+  prefill_workers: 1
+  decode_workers: 3
+  gpus_per_prefill: 2
+  gpus_per_decode: 2
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 4
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-eval-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-eval-agentic.yaml
new file mode 100644
index 0000000000..4a174c2c28
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-eval-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-1p3d-tp2-tp2-c48-fp4-eval-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: true
+  prefill_nodes: 1
+  decode_nodes: 3
+  prefill_workers: 1
+  decode_workers: 3
+  gpus_per_prefill: 2
+  gpus_per_decode: 2
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 4
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-agentic.yaml
new file mode 100644
index 0000000000..fe2d56097c
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-2p5d-tp2-tp2-c120-fp4-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: true
+  prefill_nodes: 2
+  decode_nodes: 5
+  prefill_workers: 2
+  decode_workers: 5
+  gpus_per_prefill: 2
+  gpus_per_decode: 2
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 4
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN","rejection_sample_method":"synthetic","synthetic_acceptance_length":2.78}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-eval-agentic.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-eval-agentic.yaml
new file mode 100644
index 0000000000..bf01406d14
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-eval-agentic.yaml
@@ -0,0 +1,163 @@
+name: "minimax-m3-vllm-disagg-gb300-2p5d-tp2-tp2-c120-fp4-eval-agentic"
+
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  precision: "fp4"
+
+identity:
+  model:
+    repo: "nvidia/MiniMax-M3-NVFP4"
+  container:
+    image: "vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7"
+  frameworks:
+    dynamo: "1.4.0.dev20260730"
+
+dynamo:
+  version: "1.4.0.dev20260730"
+  install: true
+
+health_check:
+  max_attempts: 2160
+  interval_seconds: 10
+
+resources:
+  gpu_type: "gb300"
+  gpus_per_node: 4
+  het_jobs: false
+  spread_workers: true
+  prefill_nodes: 2
+  decode_nodes: 5
+  prefill_workers: 2
+  decode_workers: 5
+  gpus_per_prefill: 2
+  gpus_per_decode: 2
+
+infra:
+  etcd_nats_dedicated_node: false
+  nats_max_payload_mb: 32
+
+environment:
+  PYTHONHASHSEED: "0"
+
+frontend:
+  type: dynamo
+  enable_multiple_frontends: true
+  num_additional_frontends: 4
+  nginx_session_affinity: true
+  nginx_session_affinity_header: "X-Dynamo-Session-ID"
+  args:
+    router-mode: "least-loaded"
+    router-session-affinity-ttl-secs: 1800
+  env:
+    DYN_TOKENIZER: "fastokens"
+    DYN_TOKENIZER_CACHE_BYTES: "8589934592"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+
+backend:
+  type: vllm
+  connector: null
+  dp_launch_mode: per_node
+  mooncake_kv_store:
+    env:
+      MC_ENABLE_DEST_DEVICE_AFFINITY: "1"
+      MC_STORE_CLIENT_METRIC: "1"
+      MC_STORE_CLIENT_METRIC_INTERVAL: "5"
+      MC_TE_METRIC: "0"
+    store_config:
+      metadata_server: "P2PHANDSHAKE"
+      global_segment_size: "200GB"
+      local_buffer_size: "4GB"
+      protocol: "rdma"
+      device_name: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+      mode: "embedded"
+      enable_offload: true
+  vllm_config:
+    prefill:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"recompute","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_both","kv_connector_extra_config":{"lookup_async":true}}]}}'
+      enable-cumem-allocator: true
+    decode:
+      served-model-name: "nvidia/MiniMax-M3-NVFP4"
+      kv-cache-dtype: "fp8"
+      block-size: 128
+      trust-remote-code: true
+      enable-prefix-caching: true
+      language-model-only: true
+      no-enable-flashinfer-autotune: true
+      reasoning-parser: "minimax_m3"
+      dyn-tool-call-parser: "minimax_m3"
+      dyn-reasoning-parser: "minimax_m3"
+      dyn-default-thinking-mode: "enabled"
+      max-cudagraph-capture-size: 512
+      max-num-batched-tokens: 16384
+      max-num-seqs: 1024
+      stream-interval: 20
+      gpu-memory-utilization: 0.9
+      tensor-parallel-size: 2
+      enable-expert-parallel: false
+      attention-config: '{"backend":"FLASHINFER","use_trtllm_attention":true,"indexer_kv_dtype":"fp8","minimax_m3_msa_decode_backend":"cutlass"}'
+      speculative-config: '{"method":"eagle3","model":"Inferact/MiniMax-M3-EAGLE3-GQA","num_speculative_tokens":3,"attention_backend":"FLASH_ATTN"}'
+      kv-transfer-config: '{"kv_connector":"MultiConnector","kv_role":"kv_both","kv_load_failure_policy":"fail","kv_connector_extra_config":{"connectors":[{"kv_connector":"NixlConnector","kv_role":"kv_both","kv_connector_extra_config":{"backends":["UCX"],"enforce_handshake_compat":false,"read_validation_timeout":30.0}},{"kv_connector":"MooncakeStoreConnector","kv_role":"kv_consumer","kv_connector_extra_config":{"enable_lookup":false}}]}}'
+      enable-cumem-allocator: true
+  prefill_environment: &worker_environment
+    HF_HUB_CACHE: "/hf_hub_cache"
+    HUGGINGFACE_HUB_CACHE: "/hf_hub_cache"
+    TRANSFORMERS_CACHE: "/hf_hub_cache"
+    VLLM_ENGINE_READY_TIMEOUT_S: "3600"
+    DYN_TCP_CONNECT_TIMEOUT: "120"
+    VLLM_FLOAT32_MATMUL_PRECISION: "high"
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "trtllm"
+    VLLM_USE_NCCL_SYMM_MEM: "0"
+    VLLM_ALLREDUCE_USE_SYMM_MEM: "0"
+    VLLM_MOONCAKE_LOAD_RECV_THREADS: "20"
+    UCX_MEMTYPE_CACHE: "n"
+    UCX_CUDA_IPC_ENABLE_MNNVL: "y"
+    UCX_MODULE_DIR: "/usr/local/lib/python3.12/dist-packages/nixl_cu13.libs/ucx"
+    UCX_RNDV_PIPELINE_ERROR_HANDLING: "y"
+    NCCL_CUMEM_ENABLE: "1"
+    NCCL_MNNVL_ENABLE: "1"
+    NCCL_NVLS_ENABLE: "1"
+    NCCL_IB_HCA: "mlx5_0,mlx5_1,mlx5_2,mlx5_3"
+  decode_environment:
+    <<: *worker_environment
+    VLLM_FLASHINFER_ALLREDUCE_BACKEND: "mnnvl"
+
+sbatch_directives:
+  cpus-per-task: "72"
+  mem: "0"
+
+srun_options:
+  container-remap-root: ""
+
+benchmark:
+  type: custom
+  command: "bash /infmax-workspace/benchmarks/multi_node/agentic_srt.sh"
+  env:
+    INFMAX_CONTAINER_WORKSPACE: "/infmax-workspace"
+    RESULT_DIR: "/logs/agentic"
+    PORT: "8000"
+    IS_MULTINODE: "true"
+    AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: "true"
+    AIPERF_EXTRA_INPUTS: "thinking:true"
+    AIPERF_DATASET_MMAP_CACHE_DIR: "/aiperf_mmap_cache"
+    AIPERF_SERVER_METRICS_COLLECTION_INTERVAL: "1.0"
+    HF_HUB_CACHE: "/hf_hub_cache"
+    WEKA_LOADER_OVERRIDE: "semianalysis_cc_traces_weka_062126"
diff --git a/configs/nvidia-master.yaml b/configs/nvidia-master.yaml
index 50ed9b4713..5e5cee733e 100644
--- a/configs/nvidia-master.yaml
+++ b/configs/nvidia-master.yaml
@@ -7436,6 +7436,86 @@ minimaxm3-fp4-b300-vllm-agentic-mtp:
       search-space:
       - { tp: 2, spec-decoding: mtp, kv-offloading: dram, kv-offload-backend: { name: vllm-simple }, conc-list: [24] }
 
+minimaxm3-fp4-gb300-dynamo-vllm-agentic-mtp-disagg:
+  image: vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7
+  model: nvidia/MiniMax-M3-NVFP4
+  model-prefix: minimaxm3
+  runner: cluster:gb300-nv
+  precision: fp4
+  framework: dynamo-vllm
+  router: { name: dynamo-router, version: "1.4.0.dev20260730" }
+  kv-p2p-transfer: nixl
+  multinode: true
+  disagg: true
+  scenarios:
+    agentic-coding:
+    - dram-utilization: 0.887
+      search-space:
+      - spec-decoding: mtp
+        kv-offloading: dram
+        kv-offload-backend: { name: mooncake, version: "0.3.12.post1" }
+        conc-list: [1]
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-agentic.yaml"
+          - "EVAL_CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tep4-tp4-c1-eval-agentic.yaml"
+        decode: { num-worker: 1, tp: 4, ep: 1, dp-attn: false }
+      - spec-decoding: mtp
+        kv-offloading: dram
+        kv-offload-backend: { name: mooncake, version: "0.3.12.post1" }
+        conc-list: [24]
+        prefill:
+          num-worker: 1
+          tp: 1
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-agentic.yaml"
+          - "EVAL_CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-dep4-tp4-c24-eval-agentic.yaml"
+        decode: { num-worker: 3, tp: 4, ep: 1, dp-attn: false }
+      - spec-decoding: mtp
+        kv-offloading: dram
+        kv-offload-backend: { name: mooncake, version: "0.3.12.post1" }
+        conc-list: [20, 24]
+        prefill:
+          num-worker: 1
+          tp: 2
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-agentic.yaml"
+          - "EVAL_CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p1d-tp2-tp4-c20-c24-eval-agentic.yaml"
+        decode: { num-worker: 1, tp: 4, ep: 1, dp-attn: false }
+      - spec-decoding: mtp
+        kv-offloading: dram
+        kv-offload-backend: { name: mooncake, version: "0.3.12.post1" }
+        conc-list: [48]
+        prefill:
+          num-worker: 1
+          tp: 2
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-agentic.yaml"
+          - "EVAL_CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/1p3d-tp2-tp2-c48-eval-agentic.yaml"
+        decode: { num-worker: 3, tp: 2, ep: 1, dp-attn: false }
+      - spec-decoding: mtp
+        kv-offloading: dram
+        kv-offload-backend: { name: mooncake, version: "0.3.12.post1" }
+        conc-list: [120]
+        prefill:
+          num-worker: 2
+          tp: 2
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-agentic.yaml"
+          - "EVAL_CONFIG_FILE=recipes/vllm/minimax-m3/agentic/gb300-fp4/2p5d-tp2-tp2-c120-eval-agentic.yaml"
+        decode: { num-worker: 5, tp: 2, ep: 1, dp-attn: false }
 minimaxm3-fp4-b300-trtllm-agentic-mtp:
   image: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc23.post1
   model: nvidia/MiniMax-M3-NVFP4
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index b2a5962b57..8d12d517ec 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -6263,3 +6263,14 @@
   description:
     - "Refresh with lower stream interval to collect correct client metrics"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2686
+
+- config-keys:
+    - minimaxm3-fp4-gb300-dynamo-vllm-agentic-mtp-disagg
+  scenario-type:
+    - agentic-coding
+  description:
+    - "Add the MiniMax-M3 NVFP4 GB300 disaggregated AgentX config with EAGLE3-GQA MTP and Mooncake DRAM KV offload."
+    - "Use vllm/vllm-openai:nightly-5e35a6f4f9bbc217c599692157ca985c894373f7 with golden synthetic AL 2.78 for throughput and a real-verification eval twin."
+    - "Pin thinking-on so the run matches the golden minimaxm3_eagle3_gqa thinking_on[3] = 2.78 acceptance target: dyn-default-thinking-mode enabled on the decode worker covers both the AgentX replay and the eval twin, with AIPERF_EXTRA_INPUTS thinking:true on the replay as the GB200-sibling belt-and-braces."
+    - "Set dyn-reasoning-parser and dyn-tool-call-parser alongside reasoning-parser, matching the GB200 MiniMax-M3 AgentX recipes."
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2663
diff --git a/runners/launch_gb300-nv.sh b/runners/launch_gb300-nv.sh
index ff6597d0bb..5a1d574769 100644
--- a/runners/launch_gb300-nv.sh
+++ b/runners/launch_gb300-nv.sh
@@ -80,6 +80,9 @@ elif [[ $MODEL_PREFIX == "minimaxm2.5" && $PRECISION == "fp4" ]]; then
 elif [[ $MODEL_PREFIX == "minimaxm2.5" && $PRECISION == "fp8" ]]; then
     export MODEL_PATH=/data/models/MiniMax-M2.5
     export SRT_SLURM_MODEL_PREFIX="minimax-m2.5-fp8"
+elif [[ $MODEL_PREFIX == "minimaxm3" && $PRECISION == "fp4" ]]; then
+    export MODEL_PATH=/scratch/models/MiniMax-M3-NVFP4
+    export SRT_SLURM_MODEL_PREFIX="nvidia/MiniMax-M3-NVFP4"
 elif [[ $MODEL_PREFIX == "minimaxm3" && $PRECISION == "fp8" ]]; then
     export MODEL_PATH=/data/models/MiniMax-M3-MXFP8
     export SRT_SLURM_MODEL_PREFIX="minimax-m3-mxfp8"
@@ -100,7 +103,7 @@ elif [[ $MODEL_PREFIX == "qwen3.5" && $PRECISION == "fp8" ]]; then
     export MODEL_PATH=/scratch/models/Qwen3.5-397B-A17B-FP8
     export SRT_SLURM_MODEL_PREFIX="qwen3.5-fp8"
 else
-    echo "Unsupported model: $MODEL_PREFIX-$PRECISION. Supported models are: dsr1-fp4, dsr1-fp8, dsv4-fp4, glm5-fp4, glm5-fp8, glm5.2-fp4, minimaxm2.5-fp4, minimaxm2.5-fp8, kimik2.5-fp4, kimik3-fp4, qwen3.5-fp4, qwen3.5-fp8"
+    echo "Unsupported model: $MODEL_PREFIX-$PRECISION. Supported models are: dsr1-fp4, dsr1-fp8, dsv4-fp4, glm5-fp4, glm5-fp8, glm5.2-fp4, minimaxm2.5-fp4, minimaxm2.5-fp8, minimaxm3-fp4, minimaxm3-fp8, kimik2.5-fp4, kimik3-fp4, qwen3.5-fp4, qwen3.5-fp8"
     exit 1
 fi
 
@@ -184,6 +187,10 @@ if [[ "$USES_DCGM_POWER" == "1" ]]; then
 fi
 
 export EVAL_ONLY="${EVAL_ONLY:-false}"
+if [[ "$EVAL_ONLY" == "true" && -n "${EVAL_CONFIG_FILE:-}" ]]; then
+    CONFIG_FILE="$EVAL_CONFIG_FILE"
+    echo "EVAL_ONLY=true: selecting real-verification recipe $CONFIG_FILE"
+fi
 
 export ISL="$ISL"
 export OSL="$OSL"
@@ -261,6 +268,9 @@ elif [[ "$IS_AGENTIC" == "1" ]]; then
     mkdir -p recipes/vllm/deepseek-v4/agentic || exit 1
     cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/vllm/deepseek-v4/agentic" \
         recipes/vllm/deepseek-v4/agentic || exit 1
+    mkdir -p recipes/vllm/minimax-m3/agentic || exit 1
+    cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/agentic" \
+        recipes/vllm/minimax-m3/agentic || exit 1
     mkdir -p recipes/vllm/kimi-k3/agentic || exit 1
     cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/vllm/kimi-k3/agentic" \
         recipes/vllm/kimi-k3/agentic || exit 1