diff --git a/docs/cookbook/autoregressive/DeepSeek/DeepSeek-V4_1.mdx b/docs/cookbook/autoregressive/DeepSeek/DeepSeek-V4_1.mdx
index f1f9fa46a..1b24b6d6f 100644
--- a/docs/cookbook/autoregressive/DeepSeek/DeepSeek-V4_1.mdx
+++ b/docs/cookbook/autoregressive/DeepSeek/DeepSeek-V4_1.mdx
@@ -114,6 +114,8 @@ Overriding them is the most common cause of a disappointing measurement: it leav
 
 Enable the reasoning parser — `--reasoning-parser auto` resolves to `deepseek-v41` (toggle **Reasoning Parser** in the **Parsers** card of the [Playground above](#playground)) to separate thinking from the final answer. The parser puts the thinking block in `reasoning_content` and the answer in `content`; without it both arrive concatenated in `content`.
 
+The parser can only split a thinking block the model actually produced, and **thinking is off by default** (`SGLANG_DEFAULT_THINKING=false`). Sending `reasoning_effort` turns it on: any value other than `none` also switches `chat_template_kwargs.thinking` on for you. A request that carries neither comes back with an empty `reasoning_content` however the parser is configured.
+
 <Accordion title="Reasoning Example (Python)">
 
 ```python Example
@@ -123,6 +125,7 @@ client = OpenAI(base_url="http://localhost:30000/v1", api_key="EMPTY")
 resp = client.chat.completions.create(
     model="deepseek-ai/DeepSeek-V4.1-Flash",
     messages=[{"role": "user", "content": "What is 15% of 240?"}],
+    reasoning_effort="high",
 )
 msg = resp.choices[0].message
 print("Reasoning:", getattr(msg, "reasoning_content", None))
@@ -131,7 +134,7 @@ print("Answer:", msg.content)
 
 </Accordion>
 
-Reasoning effort is part of the request contract for this model: send `reasoning_effort` on the request, either as a tier or as an integer budget. Tiers with no V4.1 counterpart (`none`, `minimal`, `medium`) log a warning and fall back to the server-side default rather than erroring; that default is `high`, and `SGLANG_DSV41_REASONING_EFFORT` overrides it.
+On the request, `reasoning_effort` accepts the tiers `low`, `high`, `xhigh` and `max`, or a float in `[0.0, 0.99]` that maps onto the model's 1–100 budget. An integer budget is reachable only through `chat_template_kwargs.reasoning_effort`, and that route does not switch thinking on by itself — pair it with `chat_template_kwargs.thinking`. `none` turns thinking off. The remaining OpenAI tiers, which have no V4.1 counterpart (`minimal`, `medium`), log a warning and fall back to the server-side default rather than erroring; that default is `high`, and `SGLANG_DSV41_REASONING_EFFORT` overrides it.
 
 ### 3.2 Tool Calling
 
@@ -160,6 +163,7 @@ resp = client.chat.completions.create(
     model="deepseek-ai/DeepSeek-V4.1-Flash",
     messages=[{"role": "user", "content": "What's the weather in Beijing?"}],
     tools=tools,
+    reasoning_effort="high",
 )
 choice = resp.choices[0]
 print("finish_reason:", choice.finish_reason)
diff --git a/docs/src/snippets/configs/deepseek-ai/deepseek-v4_1.jsx b/docs/src/snippets/configs/deepseek-ai/deepseek-v4_1.jsx
index 2985d618c..e2e1247c3 100644
--- a/docs/src/snippets/configs/deepseek-ai/deepseek-v4_1.jsx
+++ b/docs/src/snippets/configs/deepseek-ai/deepseek-v4_1.jsx
@@ -235,8 +235,8 @@ export const config = {
       ],
     },
 
-    // ---------- B200 / B300: verification round open. Mirrors the GB300 recipe
-    // because the kernels dispatch by architecture family. ----------
+    // ---------- B200: verification round open. Mirrors the GB300 recipe because
+    // the kernels dispatch by architecture family. ----------
     {
       match: { hw: "b200", strategy: "low-latency" },
       nnodes: 1,
@@ -271,10 +271,14 @@ export const config = {
         "--port {{PORT}}",
       ],
     },
+
+    // ---------- B300: 4x B300, TP4 + EP4. Same recipe as GB300 — the kernels
+    // dispatch by architecture family — with one extra memory knob on
+    // Low-Latency that GB300 does not need. ----------
     {
       match: { hw: "b300", strategy: "low-latency" },
       nnodes: 1,
-      verificationStatus: "in-progress",
+      verified: true,
       flags: [
         "--trust-remote-code",
         "--model-path {{MODEL_NAME}}",
@@ -283,6 +287,9 @@ export const config = {
         "--mem-fraction-static 0.8",
         "--speculative-algorithm DSPARK",
         "--speculative-dspark-block-size 5",
+        // GB300 does not need this, B300 does: the derived value (256) runs
+        // out of memory capturing the DSpark decode graphs at this fraction.
+        "--cuda-graph-max-bs-decode 64",
         "--reasoning-parser auto",
         "--tool-call-parser auto",
         "--host {{HOST_IP}}",
@@ -292,7 +299,7 @@ export const config = {
     {
       match: { hw: "b300", strategy: "high-throughput" },
       nnodes: 1,
-      verificationStatus: "in-progress",
+      verified: true,
       flags: [
         "--trust-remote-code",
         "--model-path {{MODEL_NAME}}",