diff --git a/docs_new/src/snippets/_playground.jsx b/docs_new/src/snippets/_playground.jsx
index c3253f935..8d4fb4ea7 100644
--- a/docs_new/src/snippets/_playground.jsx
+++ b/docs_new/src/snippets/_playground.jsx
@@ -684,6 +684,10 @@ export const Playground = ({ config }) => {
             && h.isHidden(fc.ibDevices, next.ibDevice, base)) {
           next.ibDevice = "auto"; changed = true;
         }
+        if (next.transferBackend !== "mooncake" && fc.transferBackends
+            && h.isHidden(fc.transferBackends, next.transferBackend, base)) {
+          next.transferBackend = "mooncake"; changed = true;
+        }
         return changed ? next : value;
       },
 
diff --git a/docs_new/src/snippets/configs/deepseek-ai/deepseek-v4.jsx b/docs_new/src/snippets/configs/deepseek-ai/deepseek-v4.jsx
index 28e395c86..431a47305 100644
--- a/docs_new/src/snippets/configs/deepseek-ai/deepseek-v4.jsx
+++ b/docs_new/src/snippets/configs/deepseek-ai/deepseek-v4.jsx
@@ -161,7 +161,7 @@ sgl-eval run aime25 \\
     // AMD daily-updated lmsysorg/sglang-rocm images. Bump the dated tag when you
     // re-verify on a newer build.
     mi300x: "lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi30x-20260623",
-    mi355x: "lmsysorg/sglang-rocm:v0.5.13.post1-rocm720-mi35x-20260623",
+    mi355x: "lmsysorg/sglang-rocm:v0.5.14-rocm720-mi35x-20260708",
   },
 
   // Pre-selects the issue template's `model` dropdown on "Submit verified cell".
@@ -289,6 +289,9 @@ sgl-eval run aime25 \\
           ],
           envWhen: { hw: ["gb200", "gb300"] } },
         { id: "nixl",     label: "NiXL" },
+        // MORI-IO transport is AMD-only — hidden on every non-ROCm platform.
+        { id: "mori",     label: "MORI",
+          hide: { hw: ["h100", "h200", "b200", "b300", "gb200", "gb300", "rtx6000"] } },
       ],
       // `auto` is a sentinel (emits no --disaggregation-ib-device flag).
       ibDevices: [{ id: "auto", label: "Auto" }, "mlx5_0", "mlx5_7"],
@@ -1506,7 +1509,7 @@ sgl-eval run aime25 \\
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 8192",
@@ -1523,7 +1526,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1533,12 +1539,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",
@@ -1555,7 +1560,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1565,12 +1573,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",
@@ -1599,7 +1606,7 @@ sgl-eval run aime25 \\
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 8192",
@@ -1616,7 +1623,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1626,12 +1636,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",
@@ -1648,7 +1657,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1658,12 +1670,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",
@@ -1692,7 +1703,7 @@ sgl-eval run aime25 \\
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 8192",
@@ -1709,7 +1720,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1719,12 +1733,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",
@@ -1741,7 +1754,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1751,12 +1767,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",
@@ -1785,7 +1800,7 @@ sgl-eval run aime25 \\
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 8192",
@@ -1802,7 +1817,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1812,12 +1830,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",
@@ -1834,7 +1851,10 @@ sgl-eval run aime25 \\
       verified: true,
       env: [
         "SGLANG_USE_ROCM700A=0",
+        "SGLANG_SHARED_EXPERT_TP1=1",
+        "SGLANG_DP_SHARED_EXPERT_LOCAL=1",
         "SGLANG_DP_USE_GATHERV=1",
+        "SGLANG_DP_USE_REDUCE_SCATTER=1",
         "SGLANG_HACK_FLASHMLA_BACKEND=unified_kv_triton",
         "AITER_BF16_FP8_MOE_BOUND=0",
       ],
@@ -1844,12 +1864,11 @@ sgl-eval run aime25 \\
         "--tp 8",
         "--dp 8",
         "--enable-dp-attention",
-        "--enable-prefill-delayer",
-        "--prefill-delayer-max-delay-ms 5000",
+        "--enable-two-batch-overlap",
         "--attention-backend dsv4",
         "--page-size 256",
         "--mem-fraction-static 0.90",
-        "--swa-full-tokens-ratio 0.1",
+        "--swa-full-tokens-ratio 0.15",
         "--disable-shared-experts-fusion",
         "--kv-cache-dtype fp8_e4m3",
         "--chunked-prefill-size 65536",