[XPU] Support GPT-OSS MXFP4 checkpoints on Intel XPU (#35751)

Co-authored-by: Meng, Hengyu <hengyu.meng@intel.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

docs/docs/advanced_features/quantization.mdx

@@ -47,7 +47,7 @@ The following table summarizes quantization method support across NVIDIA and AMD
       <td>Yes</td>
       <td>Yes</td>
       <td>Yes (A5)</td>
-      <td>On GPU: requires CDNA3/CDNA4 with MXFP support (uses Aiter). On Ascend NPU (A5): W4A4 MXFP4 for Qwen3 dense and MoE LLMs (MXFP4 weights + activations) — dense models support online dual-level MXFP4; offline <code>W4A4_MXFP4</code> dense and MoE checkpoints (single-level) are auto-detected via <code>modelslim</code></td>
+      <td>On GPU: requires CDNA3/CDNA4 with MXFP support (uses Aiter). On Ascend NPU (A5): W4A4 MXFP4 for Qwen3 dense and MoE LLMs (MXFP4 weights + activations) — dense models support online dual-level MXFP4; offline <code>W4A4_MXFP4</code> dense and MoE checkpoints (single-level) are auto-detected via <code>modelslim</code>. On Intel GPUs (XPU): W4A16 MoE experts on Xe2/BMG via <code>sgl-kernel-xpu</code>, enabled automatically with <code>--device xpu</code> (see <a href="../hardware-platforms/xpu">Intel GPUs (XPU)</a>)</td>
     </tr>
     <tr>
       <td><code>mxfp8</code></td>

docs/docs/hardware-platforms/xpu.mdx

@@ -41,6 +41,9 @@ A list of LLMs have been optimized on Intel GPU, and more are on the way:
 **Note:** The model identifiers listed in the table above
 have been verified on [Intel® Arc™ B580 Graphics](https://www.intel.com/content/www/us/en/products/sku/241598/intel-arc-b580-graphics/specifications.html).
 
+Quantized MoE models are covered separately in
+[MXFP4 MoE Quantization](#mxfp4-moe-quantization) below.
+
 ## Installation
 
 ### Install From Source
@@ -116,6 +119,40 @@ sglang serve                         \
     --page-size                      \   # intel_xpu attention backend supports [32, 64, 128]
 ```
 
+## MXFP4 MoE Quantization
+
+Native MXFP4 MoE checkpoints (OCP microscaling FP4: packed `e2m1` weights plus
+per-32-element `ue8m0` block scales) run on Intel GPUs through the
+`sgl-kernel-xpu` W4A16 grouped GEMM. The expert weights stay in the checkpoint's
+packed layout end to end — there is no dequantization to BF16 — so GPT-OSS-20B
+loads in roughly 13 GB rather than the ~42 GB a BF16 upcast would need, which is
+what lets it fit on a single 24 GB card.
+
+The `mxfp4` method is registered automatically on `--device xpu`, which already
+requires `sgl-kernel-xpu` — no extra flag is needed:
+
+```bash
+sglang serve                        \
+    --model-path openai/gpt-oss-20b \
+    --device xpu                    \
+    --attention-backend intel_xpu
+```
+
+**Tested models:**
+
+| Model | Notes |
+|:---|:---|
+| [openai/gpt-oss-20b](https://huggingface.co/openai/gpt-oss-20b) | 32 experts, top-k 4, hidden 2880, intermediate 2880; clamped-swiglu activation (`gemm1_alpha` 1.702, limit 7.0) with expert biases |
+
+**Requirements and limitations:**
+
+| Item | Status |
+|---|---|
+| GPU architecture | Xe2 / BMG only (Intel® Arc™ B-Series and Arc™ Pro B-Series) |
+| Quantized layers | Fused MoE experts only; attention, router, embeddings and LM head stay BF16 (per the checkpoint's `modules_to_not_convert`) |
+| Activations | BF16 (W4A16). Clamped swiglu (GPT-OSS) and plain SiLU are supported |
+| Expert parallelism | Not yet validated on this path |
+
 ## Benchmarking with Requests
 
 You can benchmark the performance via the `bench_serving` script.