[NPU] Add mxfp4-w4a4 MOE Quantization Support for NPU (#30319)

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 LLM (MXFP4 weights + activations) — online uses dual-level MXFP4, offline <code>W4A4_MXFP4</code> 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></td>
     </tr>
     <tr>
       <td><code>mxfp8</code></td>
@@ -807,6 +807,7 @@ MindStudio-ModelSlim (msModelSlim) is a model offline quantization compression t
     - [x]  ```W8A8``` linear with offline quantization of activations
     - [x]  ```W8A8_DYNAMIC``` linear with online quantization of activations
     - [x]  ```W4A4_DYNAMIC``` MOE with online quantization of activations
+    - [x]  ```W4A4_MXFP4``` MOE with dynamic MXFP4 activation quantization
     - [x]  ```W4A8_DYNAMIC``` MOE with online quantization of activations
     - [x]  ```W4A8_MXFP``` MOE with dynamic MXFP8 activation quantization
     - [x]  ```W8A8_DYNAMIC``` MOE with online quantization of activations

docs/docs/hardware-platforms/ascend-npus/optimization/quantization.mdx

@@ -84,6 +84,14 @@ SGLang supports **mix-bits** quantization (independently defines and loads each
       <td><strong style={{color: 'blue'}}>WIP</strong></td>
       <td><strong style={{color: 'red'}}>x</strong></td>
     </tr>
+    <tr>
+      <td><a href="https://github.com/sgl-project/sglang/pull/30319">MXFP4 W4A4 (ModelSlim)</a></td>
+      <td>MoE</td>
+      <td><strong style={{color: 'red'}}>x</strong></td>
+      <td><strong style={{color: 'red'}}>x</strong></td>
+      <td><strong style={{color: 'green'}}>√</strong></td>
+      <td><strong style={{color: 'red'}}>x</strong></td>
+    </tr>
     <tr>
       <td>W4A4 dynamic</td>
       <td>MoE</td>
@@ -466,6 +474,23 @@ python3 -m sglang.launch_server \
 > - As with W4A8, the packed-FP4 dtype passed to the NPU ops (`dst_type` / `x2_dtype`) must be resolved from `torch_npu.float4_e2m1fn_x2` (an int enum), not the `torch.float4_e2m1fn_x2` dtype object, which recent op-plugin builds reject.
 > - Validated end-to-end on Ascend A5 hardware.
 
+**ModelSlim W4A4 MXFP4 for LLM MoE models:**
+
+SGLang auto-detects offline ModelSlim `W4A4_MXFP4` MoE checkpoints from `quant_model_description.json`; do not pass `--quantization`. This path requires Ascend 950 products or newer.
+
+```bash Command
+MODEL_PATH=/path/to/w4a4-mxfp4-moe-model
+python3 -m sglang.launch_server \
+    --model-path "$MODEL_PATH" \
+    --device npu --attention-backend ascend \
+    --host 0.0.0.0 --port 30000 \
+    --tp-size 1
+```
+
+> **Implementation Notes:**
+> - ModelSlim supplies packed MXFP4 `w13` and `w2` expert weights with UE8M0 block scales (block size 32).
+> - Ascend TP and DeepEP dispatch activations as BF16; this path does not request MXFP4 dispatch. SGLang dynamically quantizes each expert input to MXFP4 immediately before grouped matmul.
+
 ## Diffusion Model Quantization on Ascend NPU
 
 SGLang-Diffusion supports MXFP8 online and offline quantization for diffusion models (such as Wan2.2) on Ascend NPUs. MXFP8 requires Ascend 950 Products; the ModelSlim W8A8/W4A4 schemes work on A2/A3.