diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp4-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp4-8k1k.yaml
index ad0d248260..a98e9f9c83 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp4-8k1k.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp4-8k1k.yaml
@@ -1,7 +1,7 @@
 name: "minimax-m3-vllm-disagg-b300-1p1d-tp1-tp4-fp4-8k1k"
 model:
   path: "nvidia/MiniMax-M3-NVFP4"
-  container: "vllm/vllm-openai:nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1"
+  container: "vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
   precision: "fp4"
 resources:
   gpu_type: "b300"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp8-8k1k.yaml
index 74e87048ca..5f699bbcf7 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp8-8k1k.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/1p1d-tp1-tp8-8k1k.yaml
@@ -1,7 +1,7 @@
 name: "minimax-m3-vllm-disagg-b300-1p1d-tp1-tp8-fp4-8k1k"
 model:
   path: "nvidia/MiniMax-M3-NVFP4"
-  container: "vllm/vllm-openai:nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1"
+  container: "vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
   precision: "fp4"
 resources:
   gpu_type: "b300"
@@ -27,7 +27,6 @@ backend:
   decode_environment:
     VLLM_FLOAT32_MATMUL_PRECISION: high
     UCX_TLS: "cuda_copy,rc"
-    VLLM_FLASHINFER_ALLREDUCE_BACKEND: trtllm
   vllm_config:
     prefill:
       no-enable-flashinfer-autotune: true
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/2p1d-tp1-tp4-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/2p1d-tp1-tp4-8k1k.yaml
index cb966b1e7c..7b527ce70e 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/2p1d-tp1-tp4-8k1k.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/2p1d-tp1-tp4-8k1k.yaml
@@ -1,7 +1,7 @@
 name: "minimax-m3-vllm-disagg-b300-2p1d-tp1-tp4-fp4-8k1k"
 model:
   path: "nvidia/MiniMax-M3-NVFP4"
-  container: "vllm/vllm-openai:nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1"
+  container: "vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
   precision: "fp4"
 resources:
   gpu_type: "b300"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/3p1d-tp1-tp4-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/3p1d-tp1-tp4-8k1k.yaml
index 3c6917dfa9..7bf977c39e 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/3p1d-tp1-tp4-8k1k.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/3p1d-tp1-tp4-8k1k.yaml
@@ -1,7 +1,7 @@
 name: "minimax-m3-vllm-disagg-b300-3p1d-tp1-tp4-fp4-8k1k"
 model:
   path: "nvidia/MiniMax-M3-NVFP4"
-  container: "vllm/vllm-openai:nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1"
+  container: "vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
   precision: "fp4"
 resources:
   gpu_type: "b300"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/4p1d-tp1-tep4-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/4p1d-tp1-tep4-8k1k.yaml
new file mode 100644
index 0000000000..fd11bf499f
--- /dev/null
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/4p1d-tp1-tep4-8k1k.yaml
@@ -0,0 +1,73 @@
+name: "minimax-m3-vllm-disagg-b300-4p1d-tp1-tep4-fp4-8k1k"
+model:
+  path: "nvidia/MiniMax-M3-NVFP4"
+  container: "vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
+  precision: "fp4"
+resources:
+  gpu_type: "b300"
+  gpus_per_node: 8
+  prefill_nodes: 1
+  decode_nodes: 0
+  prefill_workers: 4
+  decode_workers: 1
+  gpus_per_prefill: 1
+  gpus_per_decode: 4
+dynamo:
+  install: true
+  version: 1.3.0.dev20260713
+frontend:
+  type: dynamo
+  enable_multiple_frontends: false
+backend:
+  type: vllm
+  connector:
+  allow_prefill_decode_colocation: true
+  prefill_environment:
+    VLLM_FLOAT32_MATMUL_PRECISION: high
+    UCX_TLS: "cuda_copy,rc"
+  decode_environment:
+    VLLM_FLOAT32_MATMUL_PRECISION: high
+    UCX_TLS: "cuda_copy,rc"
+  vllm_config:
+    prefill:
+      no-enable-flashinfer-autotune: true
+      tensor-parallel-size: 1
+      enable-expert-parallel: false
+      trust-remote-code: true
+      no-enable-prefix-caching: true
+      kv-transfer-config: '{"kv_connector": "NixlConnector", "kv_role": "kv_both"}'
+      attention-config: '{"backend": "FLASHINFER", "use_trtllm_attention": true, "indexer_kv_dtype": "fp8"}'
+      kv-cache-dtype: fp8
+      block-size: 128
+      gpu-memory-utilization: 0.90
+      max-model-len: 9472
+      language-model-only: true
+      stream-interval: 32
+      enforce-eager: true
+      max-num-batched-tokens: 8192
+    decode:
+      no-enable-flashinfer-autotune: true
+      tensor-parallel-size: 4
+      enable-expert-parallel: true
+      trust-remote-code: true
+      no-enable-prefix-caching: true
+      kv-transfer-config: '{"kv_connector": "NixlConnector", "kv_role": "kv_both"}'
+      attention-config: '{"backend": "FLASHINFER", "use_trtllm_attention": true, "indexer_kv_dtype": "fp8"}'
+      kv-cache-dtype: fp8
+      block-size: 128
+      gpu-memory-utilization: 0.90
+      max-model-len: 9472
+      language-model-only: true
+      stream-interval: 32
+      max-num-seqs: 1024
+      max-num-batched-tokens: 16384
+      max-cudagraph-capture-size: 4096
+health_check:
+  max_attempts: 360
+  interval_seconds: 10
+benchmark:
+  type: "sa-bench"
+  isl: 8192
+  osl: 1024
+  concurrencies: "4096"
+  req_rate: "inf"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/5p2d-tp1-dep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/5p2d-tp1-dep8-8k1k.yaml
index 2374f1b421..55cc00b611 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/5p2d-tp1-dep8-8k1k.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/5p2d-tp1-dep8-8k1k.yaml
@@ -1,7 +1,7 @@
 name: "minimax-m3-vllm-disagg-b300-5p2d-tp1-dep8-fp4-8k1k"
 model:
   path: "nvidia/MiniMax-M3-NVFP4"
-  container: "vllm/vllm-openai:nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1"
+  container: "vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
   precision: "fp4"
 resources:
   gpu_type: "b300"
diff --git a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/8p2d-tp1-dep8-8k1k.yaml b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/8p2d-tp1-dep8-8k1k.yaml
index 88c756e071..73525d8c42 100644
--- a/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/8p2d-tp1-dep8-8k1k.yaml
+++ b/benchmarks/multi_node/srt-slurm-recipes/vllm/minimax-m3/b300-fp4/8k1k/8p2d-tp1-dep8-8k1k.yaml
@@ -1,7 +1,7 @@
 name: "minimax-m3-vllm-disagg-b300-8p2d-tp1-dep8-fp4-8k1k"
 model:
   path: "nvidia/MiniMax-M3-NVFP4"
-  container: "vllm/vllm-openai:nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1"
+  container: "vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
   precision: "fp4"
 resources:
   gpu_type: "b300"
diff --git a/configs/nvidia-master.yaml b/configs/nvidia-master.yaml
index 36ae816809..a78534ff5f 100644
--- a/configs/nvidia-master.yaml
+++ b/configs/nvidia-master.yaml
@@ -6790,7 +6790,7 @@ minimaxm3-fp8-b300-dynamo-vllm:
 # baseline above. The image includes vLLM PR #46380, so no runtime patch is
 # needed.
 minimaxm3-fp4-b300-dynamo-vllm-8k1k-tp1:
-  image: vllm/vllm-openai:nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1
+  image: vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9
   model: nvidia/MiniMax-M3-NVFP4
   model-prefix: minimaxm3
   runner: b300
@@ -6884,6 +6884,38 @@ minimaxm3-fp4-b300-dynamo-vllm-8k1k-tp1:
           ep: 8
           dp-attn: true
 
+# Compare the refreshed image on a one-node max-throughput topology: four
+# TP1 prefills colocated with one TEP4 decode worker.
+minimaxm3-fp4-b300-dynamo-vllm-8k1k-tp1-tep4:
+  image: vllm/vllm-openai:nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9
+  model: nvidia/MiniMax-M3-NVFP4
+  model-prefix: minimaxm3
+  runner: b300
+  precision: fp4
+  framework: dynamo-vllm
+  router: { name: dynamo-router, version: "1.3.0.dev20260713" }
+  kv-p2p-transfer: nixl
+  multinode: true
+  disagg: true
+  scenarios:
+    fixed-seq-len:
+    - isl: 8192
+      osl: 1024
+      search-space:
+      - conc-list: [4096]
+        prefill:
+          num-worker: 4
+          tp: 1
+          ep: 1
+          dp-attn: false
+          additional-settings:
+          - "CONFIG_FILE=recipes/vllm/minimax-m3/b300-fp4/8k1k/4p1d-tp1-tep4-8k1k.yaml"
+        decode:
+          num-worker: 1
+          tp: 4
+          ep: 4
+          dp-attn: false
+
 # Preserve the previously proven max-throughput point exactly, while keeping
 # it separate from the legacy mixed 1k1k/8k1k entry so 1k1k is not rerun.
 minimaxm3-fp4-b300-dynamo-vllm-8k1k-legacy-max-tput:
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index d969b6ce47..b34f5aa3d4 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -5129,3 +5129,12 @@
     - "Re-sweep dsv4-fp4-mi355x-atom-mtp on latest atom-dev nightly (nightly_202607231538, was atom0.1.3 release). Align ATOM official MTP config (models.json): MTP3 dp-off conc 4-256 + DPA MTP3 (dp-attn) conc 32-1024 (added c32 dp-on; tp4 tried but removed — tp4+MTP+dp-on OOMs, KV budget negative). prefill-only TBO (--enable-tbo, argparse const=prefill so enable_tbo_decode=False) on dp-attn cells at conc>=256 (measured crossover run 30257759947 vs non-TBO 30238071409: TBO -10~14% output tput at c64/c128 but +8~14% at c256+; MTP removes low-conc latency so TBO overlap there is pure overhead); MTP-safe because only decode-TBO (--enable-tbo all) drops spec_decode_metadata. max_num_seqs=conc on dp-on cells and conc>=64 to avoid OOM. bench keeps --dsv4 (InferenceX bench uses encoding_dsv4.py; DSv4-Pro has no jinja chat_template). Fixed benchmark_lib source path (../ -> ../../)."
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2345
   
+- config-keys:
+    - minimaxm3-fp4-b300-dynamo-vllm-8k1k-tp1
+    - minimaxm3-fp4-b300-dynamo-vllm-8k1k-tp1-tep4
+  description:
+    - "Update the non-speculative MiniMax-M3 NVFP4 B300 8k1k disaggregated vLLM image from nightly-2dfaae752b4db0d43cfc0715c780e33be030d0f1 to nightly-4080263bb2c5d10deac17aaeb88e0823bc35bca9"
+    - "Remove VLLM_FLASHINFER_ALLREDUCE_BACKEND=trtllm from the 1P1D TP1-prefill/TP8-decode recipe; retain the existing FP8 indexer KV cache configuration"
+    - "Add a one-node 4P1D TP1-prefill/TEP4-decode concurrency-4096 point on the refreshed image while preserving the legacy 4P2D DEP2-prefill/TEP4-decode point unchanged"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2310
+