diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx-agg.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx-agg.yaml
new file mode 100644
index 0000000000..adb7225531
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx-agg.yaml
@@ -0,0 +1,162 @@
+base:
+  name: b300-fp4-glm5.2-agentx-agg
+  model:
+    path: glm-5.2-fp4
+    container: lmsysorg/sglang:nightly-dev-cu13-20260824-95f5ecd3
+    precision: fp4
+  identity:
+    model:
+      repo: nvidia/GLM-5.2-NVFP4
+      revision: aec724e
+    frameworks:
+      sglang: nightly-dev-cu13-20260824-95f5ecd3
+  resources:
+    gpu_type: b300
+    gpus_per_node: 8
+  dynamo:
+    hash: 71eb001e17fa73c742f0afe1a6ed96836cb135fd
+    install: true
+  backend:
+    type: sglang
+    sglang_config:
+      aggregated:
+        served-model-name: GLM-5.2-FP4
+        trust-remote-code: true
+        quantization: modelopt_fp4
+        kv-cache-dtype: fp8_e4m3
+        tensor-parallel-size: 8
+        data-parallel-size: 1
+        expert-parallel-size: 1
+        enable-dp-attention: false
+        enable-dp-lm-head: false
+        max-running-requests: 10
+        cuda-graph-max-bs: 10
+        chunked-prefill-size: 8192
+        max-prefill-tokens: 8192
+        context-length: 1048576
+        speculative-algorithm: EAGLE
+        speculative-num-steps: 4
+        speculative-eagle-topk: 1
+        speculative-num-draft-tokens: 5
+        nsa-prefill-backend: trtllm
+        nsa-decode-backend: trtllm
+        moe-runner-backend: flashinfer_trtllm
+        fp4-gemm-backend: flashinfer_trtllm
+        disable-shared-experts-fusion: true
+        enable-flashinfer-allreduce-fusion: true
+        weight-loader-prefetch-checkpoints: true
+        model-loader-extra-config: '{"enable_multithread_load": true}'
+        mem-fraction-static: 0.8
+        enable-hierarchical-cache: true
+        hicache-write-policy: write_back
+        hicache-size: 270
+        hicache-io-backend: direct
+        enable-metrics: true
+        enable-cache-report: true
+    aggregated_environment:
+      TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: '1800'
+      PYTHONUNBUFFERED: '1'
+      DYN_SKIP_SGLANG_LOG_FORMATTING: '1'
+      MC_TE_METRIC: 'true'
+      NCCL_CUMEM_ENABLE: '1'
+      SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: '0'
+      SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: '1'
+      SGLANG_ENABLE_THINKING: '1'
+      SGLANG_REASONING_EFFORT: max
+      SGLANG_ENABLE_UNIFIED_RADIX_TREE: '1'
+      SGLANG_HICACHE_DEBUG_LOG: '1'
+      SGLANG_HICACHE_DEBUG_SAMPLE_RATE: '16384'
+      SGLANG_MOE_NVFP4_DISPATCH: '1'
+      SGLANG_NVFP4_CKPT_FP8_NEXTN_MOE: '1'
+      SGLANG_CLIP_MAX_NEW_TOKENS_ESTIMATION: '8'
+      SGLANG_SIMULATE_ACC_LEN: '3.33'
+      SGLANG_SIMULATE_ACC_METHOD: match-expected
+      SGLANG_SIMULATE_ACC_TOKEN_MODE: real-draft-token
+      PIP_BREAK_SYSTEM_PACKAGES: '1'
+  benchmark:
+    type: custom
+    env:
+      CONCURRENCY: '2'
+      DURATION: '3600'
+      AIPERF_HF_HOME: /hf_home
+      HF_HOME: /hf_home
+      HF_HUB_CACHE: /hf_home/hub
+      HF_DATASETS_CACHE: /hf_home/datasets
+      HF_HUB_OFFLINE: '1'
+      HF_DATASETS_OFFLINE: '1'
+      HF_HUB_DISABLE_XET: '1'
+      XDG_CACHE_HOME: /tmp/cache
+      AIPERF_DATASET_MMAP_CACHE_DIR: /hf_home/mmap_cache
+      AIPERF_DATASET_CONFIGURATION_TIMEOUT: '1800'
+      AIPERF_SERVICE_PROFILE_CONFIGURE_TIMEOUT: '1800'
+      AIPERF_FAILED_REQUEST_THRESHOLD: '0.1'
+      AIPERF_RECORD_STRIP_PAYLOAD_BYTES: 'true'
+      AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: 'true'
+      AIPERF_REQUIRED_SERVER_METRIC_PREFIX: 'sglang:'
+    command: |
+      set -euo pipefail
+      export HF_HOME="${AIPERF_HF_HOME:-/hf_home}"
+      export HF_HUB_CACHE="${HF_HOME}/hub"
+      export HF_DATASETS_CACHE="${HF_HOME}/datasets"
+      mkdir -p "${XDG_CACHE_HOME}" "${AIPERF_DATASET_MMAP_CACHE_DIR}" /logs/aiperf
+      /aiperf-venv/bin/aiperf profile \
+              --scenario inferencex-agentx-mvp \
+              --url http://localhost:8000 \
+              --endpoint /v1/chat/completions \
+              --endpoint-type chat \
+              --streaming \
+              --model GLM-5.2-FP4 \
+              --tokenizer /model \
+              --tokenizer-trust-remote-code \
+              --concurrency "${CONCURRENCY}" \
+              --benchmark-duration "${DURATION}" \
+              --random-seed 42 \
+              --failed-request-threshold "${AIPERF_FAILED_REQUEST_THRESHOLD}" \
+              --stats-interval 30 \
+              --warmup-requests-per-lane 10 \
+              --warmup-grace-period 1800 \
+              --trajectory-start-min-ratio 0.25 \
+              --trajectory-start-max-ratio 0.75 \
+              --use-server-token-count \
+              --no-gpu-telemetry \
+              --num-dataset-entries 393 \
+              --slice-duration 1.0 \
+              --public-dataset semianalysis_cc_traces_weka_062126 \
+              --trace-idle-gap-cap-seconds 300 \
+              --unsafe-override \
+              --ui simple \
+              --server-metrics-formats json \
+              --output-artifact-dir /logs/aiperf
+      rm -f /logs/aiperf/inputs.json
+  health_check:
+    max_attempts: 1440
+    interval_seconds: 10
+  infra:
+    etcd_nats_dedicated_node: false
+    nats_max_payload_mb: 64
+  frontend:
+    type: dynamo
+    env:
+      PIP_BREAK_SYSTEM_PACKAGES: '1'
+    args:
+      router-mode: kv
+      active-decode-blocks-threshold: None
+      active-prefill-tokens-threshold: None
+      active-prefill-tokens-threshold-frac: None
+      router-session-affinity-ttl-secs: 3600
+  sbatch_directives:
+    mem: '0'
+  srun_options:
+    mem: '0'
+    container-remap-root: ''
+# ################# agentx #################
+zip_override_mtp_agentx_lowlat:
+  name: [agentx-agg1_tp8-c2, agentx-agg1_tp8-c4]
+  benchmark:
+    env:
+      CONCURRENCY: ['2', '4']
+  resources:
+    agg_nodes: 1
+    agg_workers: 1
+    gpus_per_decode: 8
+    gpus_per_prefill: 8
diff --git a/benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml b/benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml
new file mode 100644
index 0000000000..d6dd47d1ae
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml
@@ -0,0 +1,233 @@
+base:
+  name: b300-fp4-glm5.2-agentx
+  model:
+    path: glm-5.2-fp4
+    container: lmsysorg/sglang:nightly-dev-cu13-20260824-95f5ecd3
+    precision: fp4
+  identity:
+    model:
+      repo: nvidia/GLM-5.2-NVFP4
+      revision: aec724e
+    frameworks:
+      dynamo: 1.2.1
+      sglang: nightly-dev-cu13-20260824-95f5ecd3
+  resources:
+    gpu_type: b300
+    gpus_per_node: 8
+  frontend:
+    type: dynamo
+    nginx_session_affinity: true
+    nginx_session_affinity_header: X-Dynamo-Session-ID
+    env:
+      PIP_BREAK_SYSTEM_PACKAGES: '1'
+    args:
+      router-mode: kv
+      router-session-affinity-ttl-secs: '3600'
+      active-decode-blocks-threshold: None
+      active-prefill-tokens-threshold: None
+      active-prefill-tokens-threshold-frac: None
+  dynamo:
+    hash: 71eb001e17fa73c742f0afe1a6ed96836cb135fd
+    install: true
+  backend:
+    type: sglang
+    prefill_environment:
+      TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: '1800'
+      PYTHONUNBUFFERED: '1'
+      DYN_SKIP_SGLANG_LOG_FORMATTING: '1'
+      SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: '100000'
+      SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: '100000'
+      SGLANG_DISAGGREGATION_WAITING_TIMEOUT: '100000'
+      MC_TE_METRIC: 'true'
+      NCCL_CUMEM_ENABLE: '1'
+      SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: '0'
+      SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: '1'
+      SGLANG_ENABLE_THINKING: '1'
+      SGLANG_ENABLE_UNIFIED_RADIX_TREE: '1'
+      SGLANG_HICACHE_DEBUG_LOG: '1'
+      SGLANG_HICACHE_DEBUG_SAMPLE_RATE: '16384'
+      SGLANG_REASONING_EFFORT: max
+      PIP_BREAK_SYSTEM_PACKAGES: '1'
+      SGLANG_SIMULATE_ACC_LEN: '2.5'
+    decode_environment:
+      TORCH_DISTRIBUTED_DEFAULT_TIMEOUT: '1800'
+      PYTHONUNBUFFERED: '1'
+      DYN_SKIP_SGLANG_LOG_FORMATTING: '1'
+      SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE: '100000'
+      SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT: '100000'
+      SGLANG_DISAGGREGATION_WAITING_TIMEOUT: '100000'
+      MC_TE_METRIC: 'true'
+      NCCL_CUMEM_ENABLE: '1'
+      SGLANG_USE_MESSAGE_QUEUE_BROADCASTER: '0'
+      SGLANG_DISABLE_TP_MEMORY_INBALANCE_CHECK: '1'
+      SGLANG_MOE_NVFP4_DISPATCH: '1'
+      SGLANG_NVFP4_CKPT_FP8_NEXTN_MOE: '1'
+      SGLANG_ENABLE_THINKING: '1'
+      SGLANG_CLIP_MAX_NEW_TOKENS_ESTIMATION: '8'
+      SGLANG_REASONING_EFFORT: max
+      PIP_BREAK_SYSTEM_PACKAGES: '1'
+      SGLANG_SIMULATE_ACC_LEN: '2.5'
+      SGLANG_SIMULATE_ACC_METHOD: match-expected
+      SGLANG_SIMULATE_ACC_TOKEN_MODE: real-draft-token
+    sglang_config:
+      prefill:
+        served-model-name: GLM-5.2-FP4
+        trust-remote-code: true
+        quantization: modelopt_fp4
+        kv-cache-dtype: fp8_e4m3
+        disaggregation-mode: prefill
+        disaggregation-transfer-backend: nixl
+        tensor-parallel-size: 4
+        data-parallel-size: 4
+        expert-parallel-size: 4
+        enable-dp-attention: true
+        enable-dp-lm-head: true
+        load-balance-method: total_tokens
+        chunked-prefill-size: 65536
+        max-prefill-tokens: 16384
+        max-running-requests: 16
+        cuda-graph-max-bs: 16
+        disable-cuda-graph: true
+        nsa-prefill-backend: trtllm
+        nsa-decode-backend: trtllm
+        moe-runner-backend: flashinfer_cutlass
+        fp4-gemm-backend: flashinfer_cutlass
+        enable-flashinfer-allreduce-fusion: true
+        weight-loader-prefetch-checkpoints: true
+        model-loader-extra-config: '{"enable_multithread_load": true}'
+        mem-fraction-static: 0.8
+        context-length: 1048576
+        enable-hierarchical-cache: true
+        hicache-write-policy: write_back
+        hicache-size: 270
+        hicache-io-backend: direct
+        speculative-algorithm: EAGLE
+        speculative-num-steps: 1
+        speculative-eagle-topk: 1
+        speculative-num-draft-tokens: 2
+        enable-metrics: true
+        enable-cache-report: true
+      decode:
+        served-model-name: GLM-5.2-FP4
+        trust-remote-code: true
+        quantization: modelopt_fp4
+        kv-cache-dtype: fp8_e4m3
+        disaggregation-mode: decode
+        disaggregation-transfer-backend: nixl
+        disable-radix-cache: true
+        speculative-algorithm: EAGLE
+        speculative-num-steps: 2
+        speculative-eagle-topk: 1
+        speculative-num-draft-tokens: 3
+        tensor-parallel-size: 4
+        data-parallel-size: 1
+        expert-parallel-size: 1
+        enable-dp-attention: false
+        enable-dp-lm-head: false
+        max-running-requests: 16
+        cuda-graph-max-bs: 16
+        chunked-prefill-size: 64
+        context-length: 1048576
+        nsa-prefill-backend: trtllm
+        nsa-decode-backend: trtllm
+        moe-runner-backend: flashinfer_trtllm
+        fp4-gemm-backend: flashinfer_cutlass
+        skip-tokenizer-init: true
+        stream-interval: 30
+        enable-flashinfer-allreduce-fusion: true
+        weight-loader-prefetch-checkpoints: true
+        model-loader-extra-config: '{"enable_multithread_load": true}'
+        mem-fraction-static: 0.92
+        disaggregation-decode-extra-slots: 0
+        enable-metrics: true
+        enable-cache-report: true
+  health_check:
+    max_attempts: 1440
+    interval_seconds: 10
+  benchmark:
+    type: custom
+    env:
+      CONCURRENCY: '63'
+      DURATION: '3600'
+      AIPERF_HF_HOME: /hf_home
+      HF_HOME: /hf_home
+      HF_HUB_CACHE: /hf_home/hub
+      HF_DATASETS_CACHE: /hf_home/datasets
+      HF_HUB_OFFLINE: '1'
+      HF_DATASETS_OFFLINE: '1'
+      HF_HUB_DISABLE_XET: '1'
+      XDG_CACHE_HOME: /tmp/cache
+      AIPERF_DATASET_MMAP_CACHE_DIR: /hf_home/mmap_cache
+      AIPERF_DATASET_CONFIGURATION_TIMEOUT: '1800'
+      AIPERF_SERVICE_PROFILE_CONFIGURE_TIMEOUT: '1800'
+      AIPERF_FAILED_REQUEST_THRESHOLD: '0.1'
+      AIPERF_RECORD_STRIP_PAYLOAD_BYTES: 'true'
+      AIPERF_HTTP_X_DYNAMO_SESSION_ID_FROM_CORRELATION_ID: 'true'
+      AIPERF_REQUIRED_SERVER_METRIC_PREFIX: 'sglang:'
+    command: |
+      set -euo pipefail
+      export HF_HOME="${AIPERF_HF_HOME:-/hf_home}"
+      export HF_HUB_CACHE="${HF_HOME}/hub"
+      export HF_DATASETS_CACHE="${HF_HOME}/datasets"
+      mkdir -p "${XDG_CACHE_HOME}" "${AIPERF_DATASET_MMAP_CACHE_DIR}" /logs/aiperf
+      /aiperf-venv/bin/aiperf profile \
+              --scenario inferencex-agentx-mvp \
+              --url http://localhost:8000 \
+              --endpoint /v1/chat/completions \
+              --endpoint-type chat \
+              --streaming \
+              --model GLM-5.2-FP4 \
+              --tokenizer /model \
+              --tokenizer-trust-remote-code \
+              --concurrency "${CONCURRENCY}" \
+              --benchmark-duration "${DURATION}" \
+              --random-seed 42 \
+              --failed-request-threshold "${AIPERF_FAILED_REQUEST_THRESHOLD}" \
+              --stats-interval 30 \
+              --warmup-requests-per-lane 10 \
+              --warmup-grace-period 1800 \
+              --trajectory-start-min-ratio 0.25 \
+              --trajectory-start-max-ratio 0.75 \
+              --use-server-token-count \
+              --no-gpu-telemetry \
+              --num-dataset-entries 393 \
+              --slice-duration 1.0 \
+              --public-dataset semianalysis_cc_traces_weka_062126 \
+              --trace-idle-gap-cap-seconds 300 \
+              --unsafe-override \
+              --ui simple \
+              --server-metrics-formats json \
+              --output-artifact-dir /logs/aiperf
+      rm -f /logs/aiperf/inputs.json
+  infra:
+    etcd_nats_dedicated_node: false
+    nats_max_payload_mb: 64
+  sbatch_directives:
+    mem: '0'
+  srun_options:
+    mem: '0'
+    container-remap-root: ''
+# ################# agentx #################
+zip_override_mtp_agentx_lowlat:
+  name: [agentx-1p2d_tp4-c63, agentx-1p3d_tp4-c61, agentx-1p4d_tp4-c59, agentx-1p6d_tp4-c58, agentx-2p1d_dtp8-c139]
+  backend:
+    sglang_config:
+      decode:
+        cuda-graph-max-bs: [16, 16, 16, 16, 139]
+        data-parallel-size: [1, 1, 1, 1, 8]
+        enable-dp-attention: [false, false, false, false, true]
+        enable-dp-lm-head: [false, false, false, false, true]
+        max-running-requests: [16, 16, 16, 16, 139]
+        tensor-parallel-size: [4, 4, 4, 4, 8]
+  benchmark:
+    env:
+      CONCURRENCY: ['63', '61', '59', '58', '139']
+  frontend:
+    enable_multiple_frontends: true
+  resources:
+    decode_nodes: [1, 2, 2, 3, 1]
+    decode_workers: [2, 3, 4, 6, 1]
+    gpus_per_decode: [4, 4, 4, 4, 8]
+    gpus_per_prefill: 4
+    prefill_nodes: 1
+    prefill_workers: [1, 1, 1, 1, 2]
diff --git a/configs/nvidia-master.yaml b/configs/nvidia-master.yaml
index 526fbec480..5804468870 100644
--- a/configs/nvidia-master.yaml
+++ b/configs/nvidia-master.yaml
@@ -9329,6 +9329,149 @@ glm5.2-fp4-gb200-dynamo-sglang-agentic-mtp-agg:
           dp-attn: false
           additional-settings:
           - CONFIG_FILE=recipes/sglang/glm5.2/gb200-fp4/agentic/agg-gb200-tp8-c8-mtp.yaml
+
+glm5.2-fp4-b300-dynamo-sglang-agentic-agg:
+  image: lmsysorg/sglang:nightly-dev-cu13-20260824-95f5ecd3
+  model: nvidia/GLM-5.2-NVFP4
+  model-prefix: glm5.2
+  runner: cluster:b300-dsxe
+  precision: fp4
+  framework: dynamo-sglang
+  router: { name: dynamo-router, version: "71eb001e17fa73c742f0afe1a6ed96836cb135fd" }
+  multinode: true
+  disagg: false
+  scenarios:
+    agentic-coding:
+    - dram-utilization: 0.80
+      search-space:
+      # aggregated 1-worker TP8 (prefill+decode on one 8-GPU B300 node), conc 2
+      - spec-decoding: mtp
+        conc-list: [2]
+        kv-offloading: dram
+        kv-offload-backend: { name: hicache }
+        num-nodes: 1
+        worker:
+          num-worker: 1
+          tp: 8
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx-agg.yaml:zip_override_mtp_agentx_lowlat[0]"
+      # aggregated 1-worker TP8 (prefill+decode on one 8-GPU B300 node), conc 4
+      - spec-decoding: mtp
+        conc-list: [4]
+        kv-offloading: dram
+        kv-offload-backend: { name: hicache }
+        num-nodes: 1
+        worker:
+          num-worker: 1
+          tp: 8
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx-agg.yaml:zip_override_mtp_agentx_lowlat[1]"
+
+glm5.2-fp4-b300-dynamo-sglang-agentic-disagg:
+  image: lmsysorg/sglang:nightly-dev-cu13-20260824-95f5ecd3
+  model: nvidia/GLM-5.2-NVFP4
+  model-prefix: glm5.2
+  runner: cluster:b300-dsxe
+  precision: fp4
+  framework: dynamo-sglang
+  router: { name: dynamo-router, version: "71eb001e17fa73c742f0afe1a6ed96836cb135fd" }
+  kv-p2p-transfer: nixl
+  multinode: true
+  disagg: true
+  scenarios:
+    agentic-coding:
+    - dram-utilization: 0.80
+      search-space:
+      # low-latency 1P2D TP4-prefill / TP4-decode (2 decode workers)
+      - spec-decoding: mtp
+        conc-list: [63]
+        kv-offloading: dram
+        kv-offload-backend: { name: hicache }
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml:zip_override_mtp_agentx_lowlat[0]"
+        decode:
+          num-worker: 2
+          tp: 4
+          ep: 1
+          dp-attn: false
+      # low-latency 1P3D TP4-prefill / TP4-decode (3 decode workers)
+      - spec-decoding: mtp
+        conc-list: [61]
+        kv-offloading: dram
+        kv-offload-backend: { name: hicache }
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml:zip_override_mtp_agentx_lowlat[1]"
+        decode:
+          num-worker: 3
+          tp: 4
+          ep: 1
+          dp-attn: false
+      # low-latency 1P4D TP4-prefill / TP4-decode (4 decode workers)
+      - spec-decoding: mtp
+        conc-list: [59]
+        kv-offloading: dram
+        kv-offload-backend: { name: hicache }
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml:zip_override_mtp_agentx_lowlat[2]"
+        decode:
+          num-worker: 4
+          tp: 4
+          ep: 1
+          dp-attn: false
+      # low-latency 1P6D TP4-prefill / TP4-decode (6 decode workers)
+      - spec-decoding: mtp
+        conc-list: [58]
+        kv-offloading: dram
+        kv-offload-backend: { name: hicache }
+        prefill:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml:zip_override_mtp_agentx_lowlat[3]"
+        decode:
+          num-worker: 6
+          tp: 4
+          ep: 1
+          dp-attn: false
+      # throughput 2P1D DEP8-decode (DP8/TP8 decode, dp-attention)
+      - spec-decoding: mtp
+        conc-list: [139]
+        kv-offloading: dram
+        kv-offload-backend: { name: hicache }
+        prefill:
+          num-worker: 2
+          tp: 4
+          ep: 4
+          dp-attn: true
+          additional-settings:
+          - "CONFIG_FILE=recipes/sglang/glm5.2/b300-fp4/agentic/glm5.2-agentx.yaml:zip_override_mtp_agentx_lowlat[4]"
+        decode:
+          num-worker: 1
+          tp: 8
+          ep: 1
+          dp-attn: true
+
 glm5.2-fp4-gb300-dynamo-sglang-agentic-agg:
   image: lmsysorg/sglang:nightly-dev-cu13-20260805-211ee642
   model: nvidia/GLM-5.2-NVFP4
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index b0dc6a14d4..0086f5e0d1 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -7145,3 +7145,14 @@
     - "Update SGLang image from lmsysorg/sglang:nightly-dev-cu13-20260907-30705c00 (2026-09-07 cu13 dev nightly, build commit sgl-project/sglang@30705c00) to the v0.5.19 release image lmsysorg/sglang:v0.5.19-cu130 (digest sha256:d6e7288627be8b02be88e4bba38e73f6d50e2826869f753c13a4c4385ab3eda9, build commit sgl-project/sglang@0bcd822377da7b5718e674eaf9c870d349424dd1, Docker Hub last pushed 2026-09-04T22:50:19Z)."
     - "The release image ships the same CUDA 13.0.3, FlashInfer 0.6.18 and sgl-kernel 0.4.6.post1 as the nightly. benchmarks/single_node/agentic/qwen3.5_fp8_h200_mtp.sh is unchanged: SGLANG_ENABLE_SPEC_V2 EAGLE MTP at 3 steps, golden acceptance length 3.39, flashinfer attention with allreduce fusion, fp8 quantization and fp8_e4m3 KV, HiCache kernel IO / page_first layout. TP8/EP1 DRAM HiCache concurrency 2 through 24 unchanged."
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2966
+
+- config-keys:
+    - glm5.2-fp4-b300-dynamo-sglang-agentic-agg
+    - glm5.2-fp4-b300-dynamo-sglang-agentic-disagg
+  scenario-type:
+    - agentic-coding
+  description:
+    - "Add GLM-5.2-NVFP4 B300 AgentX (agentic-coding) Dynamo-SGLang benchmarks over a 7-point sweep: aggregated single-node TP8 (conc 2/4), disaggregated low-latency TP4-prefill/TP4-decode 1P2D (conc 63), 1P3D (conc 61), 1P4D (conc 59), 1P6D (conc 58), and throughput 2P1D DEP8-decode (conc 139)"
+    - "EAGLE speculative decoding, hierarchical KV cache (hicache-size 270), MTP, decode mem-fraction-static 0.92, nixl KV transfer, and X-Dynamo-Session-ID session affinity; driven from the NVIDIA/srt-slurm base+zip_override recipes via CONFIG_FILE selectors"
+    - "Image: lmsysorg/sglang:nightly-dev-cu13-20260824-95f5ecd3; runner: b300-dsxe"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2654
diff --git a/runners/launch_b300-dsxe.sh b/runners/launch_b300-dsxe.sh
index 70254deb65..315d3cb4cc 100755
--- a/runners/launch_b300-dsxe.sh
+++ b/runners/launch_b300-dsxe.sh
@@ -159,27 +159,43 @@ select_srt_slurm_version() {
 SRT_REPO_DIR="srt-slurm"
 rm -rf "$SRT_REPO_DIR"
 
-if [[ "$USES_DCGM_POWER" == "1" ]]; then
-    SRT_SLURM_REPO="$POWER_SRT_SLURM_URL"
-    SRT_SLURM_REF="$POWER_SRT_SLURM_PIN"
+if [[ "$IS_AGENTIC" == "1" && $FRAMEWORK == "dynamo-sglang" && $MODEL_PREFIX == "glm5.2" && $PRECISION == "fp4" ]]; then
+    # GLM-5.2 B300 sglang AgentX: srt-slurm main carries the agentx-mvp scenario,
+    # session-affinity frontend, and custom benchmark schema these recipes need.
+    # Only glm5.2/b300-fp4 recipes get copied onto this pinned fork checkout --
+    # its recipes/ tree follows a contract other models haven't adopted yet.
+    echo "Cloning srt-slurm (https://github.com/NVIDIA/srt-slurm.git, pinned to open PR #313)..."
+    git clone https://github.com/NVIDIA/srt-slurm.git "$SRT_REPO_DIR" || exit 1
+    cd "$SRT_REPO_DIR" || exit 1
+    # NVIDIA/srt-slurm#313 is still open, so pin its reviewed recipe contract.
+    git fetch origin pull/313/head || exit 1
+    git checkout --detach 93fae852166a30f3cc054c8616228bf62c69c48c || exit 1
+    git rev-parse HEAD > "$GITHUB_WORKSPACE/srt-slurm-sha.txt"
+    mkdir -p recipes/sglang/glm5.2/b300-fp4
+    cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes/sglang/glm5.2/b300-fp4" recipes/sglang/glm5.2/b300-fp4 || exit 1
 else
-    select_srt_slurm_version
-fi
+    if [[ "$USES_DCGM_POWER" == "1" ]]; then
+        SRT_SLURM_REPO="$POWER_SRT_SLURM_URL"
+        SRT_SLURM_REF="$POWER_SRT_SLURM_PIN"
+    else
+        select_srt_slurm_version
+    fi
 
-echo "Cloning srt-slurm ($SRT_SLURM_REPO @ $SRT_SLURM_REF)..."
-git clone "$SRT_SLURM_REPO" "$SRT_REPO_DIR" || exit 1
-cd "$SRT_REPO_DIR" || exit 1
-git checkout --quiet "$SRT_SLURM_REF" || exit 1
-git rev-parse HEAD > "$GITHUB_WORKSPACE/srt-slurm-sha.txt"
-if [[ "$USES_DCGM_POWER" == "1" ]]; then
-    test "$(git rev-parse HEAD)" = "$POWER_SRT_SLURM_PIN" \
-        || { echo "Error: srt-slurm HEAD does not match POWER_SRT_SLURM_PIN=$POWER_SRT_SLURM_PIN" >&2; exit 1; }
-    cp "$GITHUB_WORKSPACE/srt-slurm-sha.txt" "$GITHUB_WORKSPACE/power-producer-sha.txt"
-fi
+    echo "Cloning srt-slurm ($SRT_SLURM_REPO @ $SRT_SLURM_REF)..."
+    git clone "$SRT_SLURM_REPO" "$SRT_REPO_DIR" || exit 1
+    cd "$SRT_REPO_DIR" || exit 1
+    git checkout --quiet "$SRT_SLURM_REF" || exit 1
+    git rev-parse HEAD > "$GITHUB_WORKSPACE/srt-slurm-sha.txt"
+    if [[ "$USES_DCGM_POWER" == "1" ]]; then
+        test "$(git rev-parse HEAD)" = "$POWER_SRT_SLURM_PIN" \
+            || { echo "Error: srt-slurm HEAD does not match POWER_SRT_SLURM_PIN=$POWER_SRT_SLURM_PIN" >&2; exit 1; }
+        cp "$GITHUB_WORKSPACE/srt-slurm-sha.txt" "$GITHUB_WORKSPACE/power-producer-sha.txt"
+    fi
 
-# Recipes live in this repo; overlay all of them onto the checkout's recipes/ dir.
-mkdir -p recipes
-cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes" recipes || exit 1
+    # Recipes live in this repo; overlay all of them onto the checkout's recipes/ dir.
+    mkdir -p recipes
+    cp -rT "$GITHUB_WORKSPACE/benchmarks/multi_node/srt-slurm-recipes" recipes || exit 1
+fi
 
 if [[ "${EVAL_FRAMEWORK:-lm-eval}" != "lm-eval" ]]; then
     python3 "$GITHUB_WORKSPACE/runners/patch_srt_eval_dispatch.py" "$(pwd)" || exit 1
@@ -293,7 +309,11 @@ SRTCTL_APPLY_ARGS=(
     --no-preflight
     --tags "b300,${MODEL_PREFIX},${PRECISION},${ISL}x${OSL},infmax-$(date +%Y%m%d)"
 )
-SRTCTL_OUTPUT=$(srtctl apply "${SRTCTL_APPLY_ARGS[@]}" 2>&1)
+if [[ "$MODEL_PREFIX" == "glm5.2" && "$FRAMEWORK" == "dynamo-sglang" ]]; then
+    SRTCTL_OUTPUT=$(env -u UCX_TLS srtctl apply "${SRTCTL_APPLY_ARGS[@]}" 2>&1)
+else
+    SRTCTL_OUTPUT=$(srtctl apply "${SRTCTL_APPLY_ARGS[@]}" 2>&1)
+fi
 echo "$SRTCTL_OUTPUT"
 
 # Extract JOB_ID from srtctl output