[Cookbook][CPU]Update CPU model support info in Cookbook (#30520)

Co-authored-by: zijiexia <37504505+zijiexia@users.noreply.github.com>

docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-OCR-2.mdx

@@ -49,7 +49,7 @@ This section provides deployment configurations optimized for different hardware
 - **Single GPU Deployment:** DeepSeek-OCR-2 (~3B parameters) fits on a single modern GPU — no tensor parallelism required for low-latency serving.
 - **High Throughput:** For high-throughput scenarios, use data parallelism with the SGLang Model Gateway. See [DP, DPA and SGLang DP Router](../../../docs/advanced_features/sgl_model_gateway).
 - **NCCL timeout:** If model loading is slow, increase `--dist-timeout 3600`.
-- For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for NUMA binding settings.
+- **Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-R1.mdx

@@ -99,7 +99,7 @@ python3 -m sglang.compile_deep_gemm --model deepseek-ai/DeepSeek-R1 --tp 8 --tru
 
 **NCCL timeout:** If model loading is slow, increase: `--dist-timeout 3600`.
 
-For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+**Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V3.mdx

@@ -79,7 +79,7 @@ DeepGEMM is enabled by default on Hopper/Blackwell and can be disabled with `SGL
 
 **NCCL timeout:** If model loading is slow and you hit an NCCL timeout, increase it: `--dist-timeout 3600`.
 
-For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+**Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V3_1.mdx

@@ -77,7 +77,7 @@ DeepGEMM is enabled by default on Hopper/Blackwell and can be disabled with `SGL
 
 **NCCL timeout:** If model loading is slow and you hit an NCCL timeout, increase it: `--dist-timeout 3600`.
 
-For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+**Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Llama/Llama3.1.mdx

@@ -22,6 +22,8 @@ SGLang offers multiple installation methods. You can choose the most suitable in
 
 Please refer to the [official SGLang installation guide](../../../docs/get-started/install) for installation instructions.
 
+For SGLang CPU installation, please refer to the [CPU version installation guide](../../../docs/hardware-platforms/cpu_server#installation).
+
 ## 3. Model Deployment
 
 This section provides deployment configurations optimized for different hardware platforms and use cases.
@@ -55,6 +57,10 @@ import { Llama31Deployment } from "/src/snippets/autoregressive/llama31-deployme
   - 70B/8B: Use AMD's optimized `amd/Llama-3.1-{size}-Instruct-FP8-KV`
 - **Tool Calling**: Enable with `--tool-call-parser llama3` for Instruct models
 
+**Xeon CPU Deployment:**
+
+- Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+
 ## 4. Model Invocation
 
 ### 4.1 Basic Usage

docs_new/cookbook/autoregressive/Llama/Llama3.3-70B.mdx

@@ -24,9 +24,11 @@ For more details, please refer to the [official Llama models repository](https:/
 
 Please refer to the [official SGLang installation guide](../../../docs/get-started/install) for installation instructions.
 
+For SGLang CPU installation, please refer to the [CPU version installation guide](../../../docs/hardware-platforms/cpu_server#installation).
+
 ## 3. Model Deployment
 
-This section provides deployment configurations optimized for AMD GPUs (MI300X, MI325X, MI355X).
+This section provides deployment configurations optimized for AMD GPUs (MI300X, MI325X, MI355X) and Intel Xeon CPUs.
 
 ### 3.1 Interactive Configuration
 
@@ -45,6 +47,10 @@ import { Llama33Deployment } from "/src/snippets/autoregressive/llama33-70b-depl
 - **Tool Calling**: Enable with `--tool-call-parser llama3` for function calling support
 - **Higher Throughput**: Optional TP=2 or TP=4 can be used for increased throughput
 
+**Xeon CPU Deployment:**
+
+Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+
 ## 4. Model Invocation
 
 ### 4.1 Basic Usage

docs_new/cookbook/autoregressive/Llama/Llama4.mdx

@@ -29,6 +29,8 @@ SGLang offers multiple installation methods. You can choose the most suitable in
 
 Please refer to the [official SGLang installation guide](../../../docs/get-started/install) for installation instructions.
 
+For SGLang CPU installation, please refer to the [CPU version installation guide](../../../docs/hardware-platforms/cpu_server#installation).
+
 ## 3. Model Deployment
 
 This section provides a progressive guide from quick deployment to performance optimization, suitable for users at different levels.
@@ -54,6 +56,7 @@ This section provides a progressive guide from quick deployment to performance o
 - **Multi-Modal:** Add `--enable-multimodal` to enable image input support.
 - **Hybrid KV Cache:** Set `--swa-full-tokens-ratio` to control the ratio of SWA (local attention) KV tokens to full-attention KV tokens (default: 0.8, range: 0–1).
 - **EAGLE Speculative Decoding:** Supported for Llama 4 Scout and Maverick via EAGLE3. Enable with the interactive command generator above.
+- **Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/OpenAI/GPT-OSS.mdx

@@ -25,6 +25,8 @@ SGLang offers multiple installation methods. You can choose the most suitable in
 
 Please refer to the [official SGLang installation guide](../../../docs/get-started/install) for installation instructions.
 
+For SGLang CPU installation, please refer to the [CPU version installation guide](../../../docs/hardware-platforms/cpu_server#installation).
+
 ## 3.Model Deployment
 
 This section provides deployment configurations optimized for different hardware platforms and use cases.
@@ -47,6 +49,7 @@ import { GPTOSSDeployment } from "/src/snippets/autoregressive/gpt-oss-deploymen
 - **MCP tool servers:** For production, point SGLang at external MCP SSE servers with `--tool-server ip-1:port-1,ip-2:port-2`.
 - **Responses API:** GPT-OSS supports OpenAI's Responses API (`client.responses.create`) in addition to the standard Chat Completions API (see section 4.2.4).
 - **Use Python 3.12** when running the demo Python tool.
+- **Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4.Model Invocation
 

docs_new/cookbook/autoregressive/Qwen/Qwen2.5-VL.mdx

@@ -29,13 +29,15 @@ SGLang offers multiple installation methods. You can choose the most suitable in
 
 Please refer to the [official SGLang installation guide](../../../docs/get-started/install) for installation instructions.
 
+For SGLang CPU installation, please refer to the [CPU version installation guide](../../../docs/hardware-platforms/cpu_server#installation).
+
 ## 3. Model Deployment
 
-This section provides deployment configurations optimized for AMD MI300X, MI325X and MI355X hardware platforms and different use cases.
+This section provides deployment configurations optimized for AMD MI300X, MI325X and MI355X as well as Intel Xeon CPU hardware platforms and different use cases.
 
 ### 3.1 Basic Configuration
 
-The Qwen2.5-VL series offers models in various sizes. The following configurations have been verified on AMD MI300X, MI325X and MI355X GPUs.
+The Qwen2.5-VL series offers models in various sizes. The following configurations have been verified on AMD MI300X, MI325X and MI355X GPUs as well as Intel Xeon CPUs.
 
 **Interactive Command Generator**: Use the configuration selector below to automatically generate the appropriate deployment command for your hardware platform and model size.
 
@@ -45,6 +47,7 @@ The Qwen2.5-VL series offers models in various sizes. The following configuratio
 
 * **Memory Management**: For the 72B model on MI300X/MI325X/MI355X, we have verified successful deployment with `--context-length 128000`. Smaller context lengths can be used to reduce memory usage if needed.
 * **Multi-GPU Deployment**: Use Tensor Parallelism (`--tp`) to scale across multiple GPUs. For example, use `--tp 8` for the 72B model and `--tp 2` for the 32B model on MI300X/MI325X/MI355X.
+* **Xeon CPU service configuration**: Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Qwen/Qwen3-Coder-Next.mdx

@@ -48,7 +48,7 @@ This section provides a progressive guide from quick deployment to performance o
 - **Mamba Radix Cache**: Qwen3-Coder-Next's hybrid Gated Delta Networks architecture supports two mamba scheduling strategies via `--mamba-scheduler-strategy`:
   - **V1 (`no_buffer`)**: Default. No overlap scheduler, lower memory usage.
   - **V2 (`extra_buffer`)**: Enables overlap scheduling and branching point caching with `--mamba-scheduler-strategy extra_buffer --page-size 64`. Requires FLA kernel backend. Trades higher mamba state memory for better throughput. Strictly superior in non-KV-cache-bound scenarios; in KV-cache-bound cases, weigh the overlap scheduling benefit against reduced max concurrency. `--page-size` must satisfy `FLA_CHUNK_SIZE % page_size == 0` or `page_size % FLA_CHUNK_SIZE == 0` (`FLA_CHUNK_SIZE` is currently 64).
-- For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+- **Xeon CPU service configuration**: Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Qwen/Qwen3-Coder.mdx

@@ -53,12 +53,12 @@ This section provides deployment configurations verified on AMD MI300X, MI325X,
 * **NVFP4 Quantization**: Requires `--quantization modelopt_fp4` and uses a different model path (`nvidia/Qwen3-Coder-...`).
 * **DP Attention**: NVFP4 configuration supports `--enable-dp-attention` for improved throughput.
 
+**Intel Xeon CPU:**
+* Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+
 **General:**
 * **Tool Use**: To enable tool calling capabilities, add `--tool-call-parser qwen3_coder` to the launch command.
 
-**Intel Xeon CPU:**
-* For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
-
 ## 4. Model Invocation
 
 ### 4.1 Basic Usage

docs_new/cookbook/autoregressive/Qwen/Qwen3-Next.mdx

@@ -58,7 +58,7 @@ The Qwen3-Next series comes in only one size but offers different thinking modes
   - **V1 (`no_buffer`)**: Default. No overlap scheduler, lower memory usage.
   - **V2 (`extra_buffer`)**: Enables overlap scheduling and branching point caching with `--mamba-scheduler-strategy extra_buffer --page-size 64`. Requires FLA kernel backend. Trades higher mamba state memory for better throughput. Strictly superior in non-KV-cache-bound scenarios; in KV-cache-bound cases, weigh the overlap scheduling benefit against reduced max concurrency. `--page-size` must satisfy `FLA_CHUNK_SIZE % page_size == 0` or `page_size % FLA_CHUNK_SIZE == 0` (`FLA_CHUNK_SIZE` is currently 64).
 
-- For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+- **Xeon CPU service configuration**: Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Qwen/Qwen3-VL.mdx

@@ -78,7 +78,7 @@ python -m sglang.launch_server \
   --enable-metrics
 ```
 
-* For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+* **Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Qwen/Qwen3.5.mdx

@@ -228,7 +228,7 @@ This section provides deployment configurations optimized for different hardware
 **FP8 KV Cache**: `--kv-cache-dtype fp8_e4m3` quantizes the KV cache to FP8 at runtime. Since these FP8 model checkpoints do not include pre-calibrated KV cache scaling factors, SGLang defaults to a scale of 1.0, which may cause noticeable accuracy degradation on reasoning-heavy tasks. It is not included in the generated commands above; add it manually only if memory constraints require the trade-off.
 </Warning>
 
-- For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+- **Xeon CPU service configuration**: Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Qwen/Qwen3.6.mdx

@@ -151,7 +151,7 @@ All Qwen3.6 variants (MoE 35B-A3B and Dense 27B) fit on a single supported GPU a
 </table>
 
 
-- For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+- **Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Qwen/Qwen3.mdx

@@ -44,16 +44,16 @@ import { Qwen3Deployment } from "/src/snippets/autoregressive/qwen3-deployment.j
 
 ### 3.2 Configuration Tips
 
-- **Memory Management** : Set lower `--context-length` to conserve memory. A value of `128000` is sufficient for most scenarios, down from the default 262K.
+- **Memory Management:** Set lower `--context-length` to conserve memory. A value of `128000` is sufficient for most scenarios, down from the default 262K.
-- **Expert Parallelism** : SGLang supports Expert Parallelism (EP) via `--ep`, allowing experts in MoE models to be deployed on separate GPUs for better throughput. One thing to note is that, for quantized models, you need to set `--ep` to a value that satisfies the requirement: `(moe_intermediate_size / moe_tp_size) % weight_block_size_n == 0, where moe_tp_size is equal to tp_size divided by ep_size.` Note that EP may perform worse in low concurrency scenarios due to additional communication overhead. Check out [Expert Parallelism Deployment](../../../docs/advanced_features/expert_parallelism) for more details.
+- **Expert Parallelism:** SGLang supports Expert Parallelism (EP) via `--ep`, allowing experts in MoE models to be deployed on separate GPUs for better throughput. One thing to note is that, for quantized models, you need to set `--ep` to a value that satisfies the requirement: `(moe_intermediate_size / moe_tp_size) % weight_block_size_n == 0, where moe_tp_size is equal to tp_size divided by ep_size.` Note that EP may perform worse in low concurrency scenarios due to additional communication overhead. Check out [Expert Parallelism Deployment](../../../docs/advanced_features/expert_parallelism) for more details.
-- **Kernel Tuning** : For MoE Triton kernel tuning on your specific hardware, refer to [fused_moe_triton](https://github.com/sgl-project/sglang/tree/main/benchmark/kernels/fused_moe_triton).
+- **Kernel Tuning:** For MoE Triton kernel tuning on your specific hardware, refer to [fused_moe_triton](https://github.com/sgl-project/sglang/tree/main/benchmark/kernels/fused_moe_triton).
-- **Speculative Decoding**: Using Speculative Decoding for latency-sensitive scenarios.
+- **Speculative Decoding:** Using Speculative Decoding for latency-sensitive scenarios.
   - `--speculative-algorithm EAGLE3`: Speculative decoding algorithm
   - `--speculative-num-steps 3`: Number of speculative verification rounds
   - `--speculative-eagle-topk 1`: Top-k sampling for draft tokens
   - `--speculative-num-draft-tokens 4`: Number of draft tokens per step
   - `--speculative-draft-model-path`: The path of the draft model weights. This can be a local folder or a Hugging Face repo ID such as [`lmsys/SGLang-EAGLE3-Qwen3-235B-A22B-Instruct-2507-SpecForge-Meituan`](https://huggingface.co/lmsys/SGLang-EAGLE3-Qwen3-235B-A22B-Instruct-2507-SpecForge-Meituan).
-- For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+- **Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/cookbook/autoregressive/Tencent/Hunyuan3-Preview.mdx

@@ -168,7 +168,7 @@ sglang serve \
 
 Toggle the "Speculative Decoding (MTP)" option in the generator above to add these flags automatically. Tune `num-steps` / `num-draft-tokens` based on acceptance rate in your workload.
 
-For configuring CPU service, please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
+**Xeon CPU service configuration:** Please refer to the `Notes` part in the serving engine launching section in [the SGLang CPU server document](../../../docs/hardware-platforms/cpu_server#launch-of-the-serving-engine) to better understand how to configure the arguments, especially for TP (tensor parallel) and NUMA binding settings.
 
 ## 4. Model Invocation
 

docs_new/src/snippets/autoregressive/gpt-oss-deployment.jsx

@@ -11,7 +11,8 @@ export const GPTOSSDeployment = () => {
         { id: 'h100', label: 'H100', default: false },
         { id: 'mi300x', label: 'MI300X', default: false },
         { id: 'mi325x', label: 'MI325X', default: false },
-        { id: 'mi355x', label: 'MI355X', default: false }
+        { id: 'mi355x', label: 'MI355X', default: false },
+        { id: 'xeon', label: 'XEON', default: false }
       ]
     },
     modelsize: {
@@ -56,6 +57,18 @@ export const GPTOSSDeployment = () => {
     }
   };
 
+  const getDisplayOptions = (values) => ({
+    ...options,
+    quantization: options.quantization,
+    speculative: {
+      ...options.speculative,
+      items: options.speculative.items.map(item => ({
+        ...item,
+        disabled: values.hardware === 'xeon' && item.id === 'enabled'
+      }))
+    }
+  });
+
   // Initialize state
   const getInitialState = () => {
     const initialState = {};
@@ -89,7 +102,13 @@ export const GPTOSSDeployment = () => {
   }, []);
 
   const handleRadioChange = (optionName, value) => {
-    setValues(prev => ({ ...prev, [optionName]: value }));
+    setValues(prev => {
+      const next = { ...prev, [optionName]: value };
+      if (optionName === 'hardware' && value === 'xeon') {
+        next.speculative = 'disabled';
+      }
+      return next;
+    });
   };
 
   // Generate command
@@ -106,7 +125,8 @@ export const GPTOSSDeployment = () => {
         b300: { tp: 8 },
         mi300x: { tp: 8 },
         mi325x: { tp: 8 },
-        mi355x: { tp: 8 }
+        mi355x: { tp: 8 },
+        xeon: { tp: 3 }
       },
       '20b': {
         baseName: '20b',
@@ -116,7 +136,8 @@ export const GPTOSSDeployment = () => {
         b300: { tp: 1 },
         mi300x: { tp: 1 },
         mi325x: { tp: 1 },
-        mi355x: { tp: 1 }
+        mi355x: { tp: 1 },
+        xeon: { tp: 3 }
       }
     };
 
@@ -146,6 +167,7 @@ export const GPTOSSDeployment = () => {
       return '# MI300X/MI325X GPUs with MXFP4 quantization: Work In Progress';
     }
 
+
     // AMD MI30x requires SGLANG_USE_AITER=0 due to YaRN RoPE precision issues
     if (hardware === 'mi300x' || hardware === 'mi325x' || hardware === 'mi355x') {
       cmd += 'SGLANG_USE_AITER=0 ';
@@ -159,6 +181,12 @@ export const GPTOSSDeployment = () => {
 
     cmd += `  --model ${modelName}`;
 
+    if (hardware === 'xeon') {
+      cmd += ` \\
+  --device cpu \\
+  --disable-overlap-schedule`;
+    }
+
     if (hwConfig.tp > 1) {
       cmd += ` \\\n  --tp ${hwConfig.tp}`;
     }
@@ -206,7 +234,7 @@ export const GPTOSSDeployment = () => {
 
   return (
     <div style={containerStyle} className="not-prose">
-      {Object.entries(options).map(([key, option]) => (
+      {Object.entries(getDisplayOptions(values)).map(([key, option]) => (
         <div key={key} style={cardStyle}>
           <div style={titleStyle}>{option.title}</div>
           <div style={itemsStyle}>
@@ -225,9 +253,10 @@ export const GPTOSSDeployment = () => {
             ) : (
               option.items.map(item => {
                 const isChecked = values[option.name] === item.id;
+                const isDisabled = Boolean(item.disabled);
                 return (
-                  <label key={item.id} style={{ ...labelBaseStyle, ...(isChecked ? checkedStyle : {}) }}>
+                  <label key={item.id} style={{ ...labelBaseStyle, ...(isChecked ? checkedStyle : {}), ...(isDisabled ? disabledStyle : {}) }}>
-                    <input type="radio" name={option.name} value={item.id} checked={isChecked} onChange={() => handleRadioChange(option.name, item.id)} style={{ display: 'none' }} />
+                    <input type="radio" name={option.name} value={item.id} checked={isChecked} disabled={isDisabled} onChange={() => !isDisabled && handleRadioChange(option.name, item.id)} style={{ display: 'none' }} />
                     {item.label}
                     {item.subtitle && <small style={{ ...subtitleStyle, color: isChecked ? 'rgba(255,255,255,0.85)' : 'inherit' }}>{item.subtitle}</small>}
                   </label>

docs_new/src/snippets/autoregressive/llama31-deployment.jsx

@@ -10,7 +10,8 @@ export const Llama31Deployment = () => {
         { id: 'b200', label: 'B200', default: false },
         { id: 'mi300x', label: 'MI300X', default: false },
         { id: 'mi325x', label: 'MI325X', default: false },
-        { id: 'mi355x', label: 'MI355X', default: false }
+        { id: 'mi355x', label: 'MI355X', default: false },
+        { id: 'xeon', label: 'XEON', default: false }
       ]
     },
     modelsize: {
@@ -57,6 +58,34 @@ export const Llama31Deployment = () => {
     }
   };
 
+  const getDisplayOptions = (values) => {
+    const displayOptions = {
+      ...options,
+      modelsize: {
+        ...options.modelsize,
+        items: options.modelsize.items.map(item => ({
+          ...item,
+          disabled: values.hardware === 'xeon' && item.id !== '8b'
+        }))
+      },
+      quantization: {
+        ...options.quantization,
+        items: options.quantization.items.map(item => ({
+          ...item,
+          disabled: values.hardware === 'xeon' && item.id === 'fp8'
+        }))
+      },
+      optimization: {
+        ...options.optimization,
+        items: options.optimization.items.map(item => ({
+          ...item,
+          disabled: values.hardware === 'xeon' && item.id !== 'basic'
+        }))
+      }
+    };
+    return displayOptions;
+  };
+
   // Initialize state
   const getInitialState = () => {
     const initialState = {};
@@ -86,7 +115,15 @@ export const Llama31Deployment = () => {
   }, []);
 
   const handleRadioChange = (optionName, value) => {
-    setValues(prev => ({ ...prev, [optionName]: value }));
+    setValues(prev => {
+      const next = { ...prev, [optionName]: value };
+      if (optionName === 'hardware' && value === 'xeon') {
+        next.modelsize = '8b';
+        next.quantization = 'bf16';
+        next.optimization = 'basic';
+      }
+      return next;
+    });
   };
 
   // Generate command
@@ -94,6 +131,8 @@ export const Llama31Deployment = () => {
     const { hardware, optimization, modelsize, category, toolcall, quantization } = values;
 
     const isAMD = hardware === 'mi300x' || hardware === 'mi325x' || hardware === 'mi355x';
+    const isXeon = hardware === 'xeon';
+    const effectiveModelSize = isXeon ? '8b' : modelsize;
 
     // Model size mapping
     const sizeMap = {
@@ -101,13 +140,13 @@ export const Llama31Deployment = () => {
       '70b': '70B',
       '405b': '405B'
     };
-    const sizeToken = sizeMap[modelsize] || '70B';
+    const sizeToken = sizeMap[effectiveModelSize] || '70B';
     const categorySuffix = category === 'instruct' ? '-Instruct' : '';
 
     // Determine model path
     let modelPath;
-    if (quantization === 'fp8' && category === 'instruct') {
+    if (quantization === 'fp8' && category === 'instruct' && !isXeon) {
-      if (modelsize === '405b') {
+      if (effectiveModelSize === '405b') {
         // Meta official FP8 for 405B
         modelPath = `meta-llama/Llama-3.1-${sizeToken}${categorySuffix}-FP8`;
       } else if (isAMD) {
@@ -142,13 +181,16 @@ export const Llama31Deployment = () => {
         }
       };
       tpSize = quantization === 'fp8'
-        ? amdTpConfig[hardware][modelsize].fp8
+        ? amdTpConfig[hardware][effectiveModelSize].fp8
-        : amdTpConfig[hardware][modelsize].bf16;
+        : amdTpConfig[hardware][effectiveModelSize].bf16;
+    } else if (isXeon) {
+      // Intel Xeon CPU TP configuration
+      tpSize = 3;
     } else {
       // NVIDIA GPU TP configuration
-      if (modelsize === '405b') {
+      if (effectiveModelSize === '405b') {
         tpSize = 8;
-      } else if (modelsize === '70b' && (hardware === 'h100' || hardware === 'h200')) {
+      } else if (effectiveModelSize === '70b' && (hardware === 'h100' || hardware === 'h200')) {
         tpSize = 2;
       }
     }
@@ -157,17 +199,22 @@ export const Llama31Deployment = () => {
     const args = [];
     args.push(`--model-path ${modelPath}`);
 
+    if (isXeon) {
+      args.push(`--device cpu`);
+      args.push(`--disable-overlap-schedule`);
+    }
+
     if (tpSize) {
       args.push(`--tp ${tpSize}`);
     }
 
     // Add quantization flag only if not using FP8 variant model
-    if (quantization === 'fp8' && category !== 'instruct') {
+    if (quantization === 'fp8' && category !== 'instruct' && !isXeon) {
       args.push(`--quantization fp8`);
     }
 
     // NVIDIA-specific optimizations
-    if (!isAMD) {
+    if (!isAMD && !isXeon) {
       if (optimization === 'throughput') {
         args.push(`--enable-dp-attention`);
         args.push(`--mem-fraction-static 0.85`);
@@ -176,7 +223,7 @@ export const Llama31Deployment = () => {
         args.push(`--speculative-num-steps 3`);
         args.push(`--speculative-eagle-topk 1`);
         args.push(`--speculative-num-draft-tokens 4`);
-        if (modelsize === '8b' && category === 'instruct') {
+        if (effectiveModelSize === '8b' && category === 'instruct') {
           args.push(`--speculative-draft-model-path yuhuili/EAGLE3-LLaMA3.1-Instruct-8B`);
         } else {
           args.push(`--speculative-draft-model-path \${EAGLE3_MODEL_PATH}`);
@@ -212,7 +259,7 @@ export const Llama31Deployment = () => {
 
   return (
     <div style={containerStyle} className="not-prose">
-      {Object.entries(options).map(([key, option]) => (
+      {Object.entries(getDisplayOptions(values)).map(([key, option]) => (
         <div key={key} style={cardStyle}>
           <div style={titleStyle}>{option.title}</div>
           <div style={itemsStyle}>
@@ -231,9 +278,10 @@ export const Llama31Deployment = () => {
             ) : (
               option.items.map(item => {
                 const isChecked = values[option.name] === item.id;
+                const isDisabled = Boolean(item.disabled);
                 return (
-                  <label key={item.id} style={{ ...labelBaseStyle, ...(isChecked ? checkedStyle : {}) }}>
+                  <label key={item.id} style={{ ...labelBaseStyle, ...(isChecked ? checkedStyle : {}), ...(isDisabled ? disabledStyle : {}) }}>
-                    <input type="radio" name={option.name} value={item.id} checked={isChecked} onChange={() => handleRadioChange(option.name, item.id)} style={{ display: 'none' }} />
+                    <input type="radio" name={option.name} value={item.id} checked={isChecked} disabled={isDisabled} onChange={() => !isDisabled && handleRadioChange(option.name, item.id)} style={{ display: 'none' }} />
                     {item.label}
                     {item.subtitle && <small style={{ ...subtitleStyle, color: isChecked ? 'rgba(255,255,255,0.85)' : 'inherit' }}>{item.subtitle}</small>}
                   </label>

docs_new/src/snippets/autoregressive/llama33-70b-deployment.jsx

@@ -7,7 +7,8 @@ export const Llama33Deployment = () => {
       items: [
         { id: 'mi300x', label: 'MI300X', default: true },
         { id: 'mi325x', label: 'MI325X', default: false },
-        { id: 'mi355x', label: 'MI355X', default: false }
+        { id: 'mi355x', label: 'MI355X', default: false },
+        { id: 'xeon', label: 'XEON', default: false }
       ]
     },
     quantization: {
@@ -28,6 +29,17 @@ export const Llama33Deployment = () => {
     }
   };
 
+  const getDisplayOptions = (values) => ({
+    ...options,
+    quantization: {
+      ...options.quantization,
+      items: options.quantization.items.map(item => ({
+        ...item,
+        disabled: values.hardware === 'xeon' && item.id === 'fp8'
+      }))
+    }
+  });
+
   // Initialize state
   const getInitialState = () => {
     const initialState = {};
@@ -57,7 +69,13 @@ export const Llama33Deployment = () => {
   }, []);
 
   const handleRadioChange = (optionName, value) => {
-    setValues(prev => ({ ...prev, [optionName]: value }));
+    setValues(prev => {
+      const next = { ...prev, [optionName]: value };
+      if (optionName === 'hardware' && value === 'xeon') {
+        next.quantization = 'bf16';
+      }
+      return next;
+    });
   };
 
   // Generate command
@@ -65,14 +83,20 @@ export const Llama33Deployment = () => {
     const { hardware, quantization, toolcall } = values;
 
     // Select model based on quantization
-    const modelPath = quantization === 'fp8'
+    const modelPath = quantization === 'fp8' && hardware !== 'xeon'
       ? 'amd/Llama-3.3-70B-Instruct-FP8-KV'
       : 'meta-llama/Llama-3.3-70B-Instruct';
 
     // Build command
     let cmd = 'python -m sglang.launch_server \\\n';
     cmd += `  --model-path ${modelPath} \\\n`;
+    if (hardware === 'xeon') {
+      cmd += `  --device cpu \\\n`;
+      cmd += `  --disable-overlap-schedule \\\n`;
+      cmd += `  --tp 6`;
+    } else {
       cmd += `  --tp 1`;
+    }
 
     // Add tool calling parser
     if (toolcall === 'enabled') {
@@ -98,7 +122,7 @@ export const Llama33Deployment = () => {
 
   return (
     <div style={containerStyle} className="not-prose">
-      {Object.entries(options).map(([key, option]) => (
+      {Object.entries(getDisplayOptions(values)).map(([key, option]) => (
         <div key={key} style={cardStyle}>
           <div style={titleStyle}>{option.title}</div>
           <div style={itemsStyle}>
@@ -117,9 +141,10 @@ export const Llama33Deployment = () => {
             ) : (
               option.items.map(item => {
                 const isChecked = values[option.name] === item.id;
+                const isDisabled = Boolean(item.disabled);
                 return (
-                  <label key={item.id} style={{ ...labelBaseStyle, ...(isChecked ? checkedStyle : {}) }}>
+                  <label key={item.id} style={{ ...labelBaseStyle, ...(isChecked ? checkedStyle : {}), ...(isDisabled ? disabledStyle : {}) }}>
-                    <input type="radio" name={option.name} value={item.id} checked={isChecked} onChange={() => handleRadioChange(option.name, item.id)} style={{ display: 'none' }} />
+                    <input type="radio" name={option.name} value={item.id} checked={isChecked} disabled={isDisabled} onChange={() => !isDisabled && handleRadioChange(option.name, item.id)} style={{ display: 'none' }} />
                     {item.label}
                     {item.subtitle && <small style={{ ...subtitleStyle, color: isChecked ? 'rgba(255,255,255,0.85)' : 'inherit' }}>{item.subtitle}</small>}
                   </label>

docs_new/src/snippets/autoregressive/llama4-maverick-deployment.jsx

@@ -8,15 +8,16 @@ export const Llama4MaverickDeployment = () => {
         { id: 'h200', label: 'H200', default: false },
         { id: 'mi300x', label: 'MI300x', default: true },
         { id: 'mi325x', label: 'MI325x', default: false },
-        { id: 'mi355x', label: 'MI355x', default: false }
+        { id: 'mi355x', label: 'MI355x', default: false },
+        { id: 'xeon', label: 'XEON', default: false }
       ]
     },
     quantization: {
       name: 'quantization',
       title: 'Quantization',
-      items: [
+      getDynamicItems: (values) => [
         { id: 'bf16', label: 'BF16', default: true },
-        { id: 'fp8', label: 'FP8', default: false }
+        { id: 'fp8', label: 'FP8', default: false, disabled: values.hardware === 'xeon' }
       ]
     },
     toolcall: {
@@ -30,6 +31,7 @@ export const Llama4MaverickDeployment = () => {
     speculative: {
       name: 'speculative',
       title: 'Speculative Decoding (EAGLE3)',
+      condition: (values) => values.hardware !== 'xeon',
       items: [
         { id: 'disabled', label: 'Disabled', default: true },
         { id: 'enabled', label: 'Enable EAGLE3', default: false }
@@ -63,9 +65,11 @@ export const Llama4MaverickDeployment = () => {
       cmd += ` \\\n  --tp 8`;
     } else if (hardware === 'mi300x' || hardware === 'mi325x' || hardware === 'mi355x') {
       cmd += ` \\\n  --tp 8`;
+    } else if (hardware === 'xeon') {
+      cmd += ` \\\n  --device cpu \\\n  --disable-overlap-schedule \\\n  --tp 6`;
     }
 
-    if (quantization === 'fp8') {
+    if (quantization === 'fp8' && hardware !== 'xeon') {
       cmd += ` \\\n  --quantization fp8`;
     }
 
@@ -73,7 +77,7 @@ export const Llama4MaverickDeployment = () => {
       cmd += ` \\\n  --tool-call-parser pythonic`;
     }
 
-    if (speculative === 'enabled') {
+    if (speculative === 'enabled' && hardware !== 'xeon') {
       cmd += ` \\\n  --speculative-algorithm EAGLE3 \\\n`;
       cmd += `  --speculative-draft-model-path lmsys/sglang-EAGLE3-Llama-4-Maverick-17B-128E-Instruct-v1 \\\n`;
       cmd += `  --speculative-num-steps 3 \\\n`;
@@ -151,7 +155,14 @@ export const Llama4MaverickDeployment = () => {
   }, []);
 
   const handleRadioChange = (optionName, value) => {
-    setValues((prev) => ({ ...prev, [optionName]: value }));
+    setValues((prev) => {
+      const next = { ...prev, [optionName]: value };
+      if (optionName === 'hardware' && value === 'xeon') {
+        next.quantization = 'bf16';
+        next.speculative = 'disabled';
+      }
+      return next;
+    });
   };
 
   const handleCheckboxChange = (optionName, itemId, isChecked) => {

docs_new/src/snippets/autoregressive/llama4-scout-deployment.jsx

@@ -9,15 +9,16 @@ export const Llama4ScoutDeployment = () => {
         { id: 'h200', label: 'H200', default: false },
         { id: 'mi300x', label: 'MI300x', default: false },
         { id: 'mi325x', label: 'MI325x', default: false },
-        { id: 'mi355x', label: 'MI355x', default: false }
+        { id: 'mi355x', label: 'MI355x', default: false },
+        { id: 'xeon', label: 'XEON', default: false }
       ]
     },
     quantization: {
       name: 'quantization',
       title: 'Quantization',
-      items: [
+      getDynamicItems: (values) => [
         { id: 'bf16', label: 'BF16', default: true },
-        { id: 'fp8', label: 'FP8', default: false }
+        { id: 'fp8', label: 'FP8', default: false, disabled: values.hardware === 'xeon' }
       ]
     },
     toolcall: {
@@ -31,6 +32,7 @@ export const Llama4ScoutDeployment = () => {
     speculative: {
       name: 'speculative',
       title: 'Speculative Decoding (EAGLE3)',
+      condition: (values) => values.hardware !== 'xeon',
       items: [
         { id: 'disabled', label: 'Disabled', default: true },
         { id: 'enabled', label: 'Enable EAGLE3', default: false }
@@ -64,9 +66,11 @@ export const Llama4ScoutDeployment = () => {
       cmd += ` \\\n  --tp 8`;
     } else if (hardware === 'mi300x' || hardware === 'mi325x' || hardware === 'mi355x') {
       cmd += ` \\\n  --tp 8`;
+    } else if (hardware === 'xeon') {
+      cmd += ` \\\n  --device cpu \\\n  --disable-overlap-schedule \\\n  --tp 6`;
     }
 
-    if (quantization === 'fp8') {
+    if (quantization === 'fp8' && hardware !== 'xeon') {
       cmd += ` \\\n  --quantization fp8`;
     }
 
@@ -74,7 +78,7 @@ export const Llama4ScoutDeployment = () => {
       cmd += ` \\\n  --tool-call-parser pythonic`;
     }
 
-    if (speculative === 'enabled') {
+    if (speculative === 'enabled' && hardware !== 'xeon') {
       cmd += ` \\\n  --speculative-algorithm EAGLE3 \\\n`;
       cmd += `  --speculative-draft-model-path lmsys/sglang-EAGLE3-Llama-4-Scout-17B-16E-Instruct-v1 \\\n`;
       cmd += `  --speculative-num-steps 3 \\\n`;
@@ -152,7 +156,14 @@ export const Llama4ScoutDeployment = () => {
   }, []);
 
   const handleRadioChange = (optionName, value) => {
-    setValues((prev) => ({ ...prev, [optionName]: value }));
+    setValues((prev) => {
+      const next = { ...prev, [optionName]: value };
+      if (optionName === 'hardware' && value === 'xeon') {
+        next.quantization = 'bf16';
+        next.speculative = 'disabled';
+      }
+      return next;
+    });
   };
 
   const handleCheckboxChange = (optionName, itemId, isChecked) => {

docs_new/src/snippets/autoregressive/qwen25-vl-deployment.jsx

@@ -6,7 +6,8 @@ export const Qwen25VLDeployment = () => {
       items: [
         { id: 'mi300x', label: 'MI300X', default: true },
         { id: 'mi325x', label: 'MI325X', default: false },
-        { id: 'mi355x', label: 'MI355X', default: false }
+        { id: 'mi355x', label: 'MI355X', default: false },
+        { id: 'xeon', label: 'XEON', default: false }
       ]
     },
     modelsize: {
@@ -33,25 +34,29 @@ export const Qwen25VLDeployment = () => {
       baseName: '72B',
       mi300x: { tp: 8, ep: 0 },
       mi325x: { tp: 8, ep: 0 },
-      mi355x: { tp: 8, ep: 0 }
+      mi355x: { tp: 8, ep: 0 },
+      xeon: { tp: 6, ep: 0 }
     },
     '32b': {
       baseName: '32B',
       mi300x: { tp: 2, ep: 0 },
       mi325x: { tp: 2, ep: 0 },
-      mi355x: { tp: 2, ep: 0 }
+      mi355x: { tp: 2, ep: 0 },
+      xeon: { tp: 6, ep: 0 }
     },
     '7b': {
       baseName: '7B',
       mi300x: { tp: 1, ep: 0 },
       mi325x: { tp: 1, ep: 0 },
-      mi355x: { tp: 1, ep: 0 }
+      mi355x: { tp: 1, ep: 0 },
+      xeon: { tp: 3, ep: 0 }
     },
     '3b': {
       baseName: '3B',
       mi300x: { tp: 1, ep: 0 },
       mi325x: { tp: 1, ep: 0 },
-      mi355x: { tp: 1, ep: 0 }
+      mi355x: { tp: 1, ep: 0 },
+      xeon: { tp: 3, ep: 0 }
     }
   };
 
@@ -73,6 +78,10 @@ export const Qwen25VLDeployment = () => {
     let cmd = 'python -m sglang.launch_server \\\n';
     cmd += `  --model ${modelName}`;
 
+    if (hardware === 'xeon') {
+      cmd += ` \\\n  --device cpu \\\n  --disable-overlap-schedule`;
+    }
+
     if (hwConfig.tp > 1) {
       cmd += ` \\\n  --tp ${hwConfig.tp}`;
     }