[Cookbook] Enable DSpark on the DeepSeek-V4 Flash Vision low-latency recipes (#37301)

docs/cookbook/autoregressive/DeepSeek/DeepSeek-V4.mdx

@@ -225,7 +225,7 @@ For the original Flash and Pro checkpoints:
 - **Verified matrix** — MMMU-Pro via sgl-eval at `temperature 1.0`, `top-p 0.95`, `--reasoning-effort max`.
 - **Engine auto-configuration** — the engine picks the `flashinfer_mxfp4` MoE runner and auto-disables shared-experts fusion for this checkpoint (its HashTopK routing rejects fused shared experts); don't pass `--enforce-shared-experts-fusion`.
 - **Chunked prefill & radix cache stay enabled** — the scheduler keeps image spans consistent automatically: chunked-prefill truncation points are span-aligned (an image span always prefills within a single extend, overshooting the chunk budget by at most one span), and a radix-cache prefix match ending deep inside an image span is re-issued from the span start.
-- **Speculative decoding** — the checkpoint bundles a DSpark head, but MTP/DSpark with image batches is not yet verified, so all Flash Vision recipes run target-only. As on the 0731/0813 checkpoints, do not pass the EAGLE flags.
+- **Speculative decoding** — the checkpoint bundles a DSpark head, and the low-latency recipes enable it with `--speculative-algorithm DSPARK` (verified with image batches on B200 via the MMMU-Pro round; the other hardware rows are pending verification). The balanced and high-throughput recipes run target-only: they use DP Attention, which DSpark is incompatible with on current releases. As on the 0731/0813 checkpoints, do not pass the EAGLE flags.
 
 **Shared experts fusion (Blackwell, flashinfer_mxfp4)**
 
@@ -625,7 +625,7 @@ For more details, see the [HiCache documentation](../../../docs/advanced_feature
 
 Flash Official (0731) and Pro Official (0813) bundle a DSpark draft head in `deepseek-ai/DeepSeek-V4-Flash-0731` and `deepseek-ai/DeepSeek-V4-Pro-0813`. The target and draft weights therefore come from the same checkpoint: enable DSpark with `--speculative-algorithm DSPARK` and do not set a separate `--speculative-draft-model-path`.
 
-The experimental [Flash Vision checkpoint](#vision-note) also bundles a DSpark head, but speculative decoding with image batches is not yet verified on it — the Flash Vision recipes run target-only for now, and the Playground greys the DSpark chip out on that variant.
+The experimental [Flash Vision checkpoint](#vision-note) also bundles a DSpark head, enabled the same way: the Flash Vision low-latency recipes ship with `--speculative-algorithm DSPARK` (verified with image batches on B200 via the MMMU-Pro round; other hardware rows pending). The balanced and high-throughput Flash Vision recipes stay target-only because they run DP Attention.
 
 Unlike the EAGLE recipes for the original Flash and Pro checkpoints, this recipe omits `--speculative-num-steps`, `--speculative-eagle-topk`, and `--speculative-num-draft-tokens`. SGLang reads the DSpark shape from the checkpoint.
 

docs/src/snippets/configs/deepseek-ai/deepseek-v4-benchmarks.jsx

@@ -640,8 +640,8 @@ export const benchmarks = [
   {
     match: { hw: "b200", variant: "flash-vision", quant: "fp4", strategy: "low-latency", nodes: "single" },
     sglang_version: "dev-dsv4-flash-vision",
-    accuracy: { mmmu_pro_pct: 74.96 },
+    accuracy: { mmmu_pro_pct: 75.14 },
-    notes: "MMMU-Pro (standard, 10-option) measured with sgl-eval on 4×B200 (TP=4) at temperature 1.0, top-p 0.95, --reasoning-effort max.",
+    notes: "MMMU-Pro (standard, 10-option) measured with sgl-eval on 4×B200 (TP=4) at temperature 1.0, top-p 0.95, --reasoning-effort max, with the bundled DSpark head enabled (--speculative-algorithm DSPARK).",
   },
   { match: { hw: "b200", variant: "flash-vision", quant: "fp4", strategy: "balanced", nodes: "single" } },
   { match: { hw: "b200", variant: "flash-vision", quant: "fp4", strategy: "high-throughput", nodes: "single" } },

docs/src/snippets/configs/deepseek-ai/deepseek-v4.jsx

@@ -313,8 +313,6 @@ sgl-eval run mmmu_pro \\
           flags: ["--speculative-algorithm DSPARK"],
           hide: { variant: ["flash", "pro"] },
           disable: [
-            { when: { variant: ["flash-vision"] },
-              reason: "The Flash Vision checkpoint bundles a DSpark head, but speculative decoding is not yet verified with image inputs — the cookbook recipes run target-only for now." },
             { when: { dpAttnOn: [true] },
               reason: "DSpark is not compatible with DP Attention on the current release." },
             { when: { hw: ["mi300x", "mi355x"] },
@@ -2849,11 +2847,12 @@ sgl-eval run mmmu_pro \\
     //
     // DeepSeek-V4-Flash-Vision-Exp (sgl-project/sglang#37253): the 0731
     // Flash base plus a vision encoder + aligner. The checkpoint bundles a
-    // DSpark head, but speculative decoding is not yet verified with image
+    // DSpark head; low-latency recipes enable it (--speculative-algorithm
-    // batches, so every recipe runs target-only. Low-latency is the serving
+    // DSPARK, no other spec flags — the draft ships in the main checkpoint),
-    // shape the MMMU-Pro round ran on (4×B200); balanced / high-throughput
+    // verified on B200 via the MMMU-Pro round (4×B200, image batches).
-    // mirror the Flash Official recipes on the same 4-GPU topology — final
+    // Balanced / high-throughput stay target-only: those recipes run DP
-    // verification in progress.
+    // attention, which DSpark is incompatible with on the current release.
+    // Non-B200 hardware — final verification in progress.
     // ====================================================================
     {
       match: { hw: "b200", variant: "flash-vision", quant: "fp4", strategy: "low-latency", nodes: "single" },
@@ -2863,6 +2862,7 @@ sgl-eval run mmmu_pro \\
       flags: [
         "--model-path {{MODEL_NAME}}",
         "--tp 4",
+        "--speculative-algorithm DSPARK",
         "--mem-fraction-static 0.85",
         "--host {{HOST_IP}}",
         "--port {{PORT}}",
@@ -2918,6 +2918,7 @@ sgl-eval run mmmu_pro \\
       flags: [
         "--model-path {{MODEL_NAME}}",
         "--tp 4",
+        "--speculative-algorithm DSPARK",
         "--mem-fraction-static 0.85",
         "--host {{HOST_IP}}",
         "--port {{PORT}}",
@@ -2969,6 +2970,7 @@ sgl-eval run mmmu_pro \\
       flags: [
         "--model-path {{MODEL_NAME}}",
         "--tp 4",
+        "--speculative-algorithm DSPARK",
         "--mem-fraction-static 0.85",
         "--host {{HOST_IP}}",
         "--port {{PORT}}",
@@ -3020,6 +3022,7 @@ sgl-eval run mmmu_pro \\
       flags: [
         "--model-path {{MODEL_NAME}}",
         "--tp 4",
+        "--speculative-algorithm DSPARK",
         "--mem-fraction-static 0.85",
         "--host {{HOST_IP}}",
         "--port {{PORT}}",
@@ -3076,6 +3079,7 @@ sgl-eval run mmmu_pro \\
         "--model-path {{MODEL_NAME}}",
         "--tp 4",
         "--moe-runner-backend marlin",
+        "--speculative-algorithm DSPARK",
         "--mem-fraction-static 0.85",
         "--host {{HOST_IP}}",
         "--port {{PORT}}",