diff --git a/docs/compute/clusters_gpu/iluvatar_gpu.md b/docs/compute/clusters_gpu/iluvatar_gpu.md
index 0339df45a704e432bbb07073923aff7a185e073d..0a5f7ecabee4b96c46ef6541c748a7bfdededf3a 100644
--- a/docs/compute/clusters_gpu/iluvatar_gpu.md
+++ b/docs/compute/clusters_gpu/iluvatar_gpu.md
@@ -61,16 +61,16 @@ sidebar_position: 3
 
 ## 4. AI 开发库：适配与版本管理
 
-为了发挥智铠100 的最佳性能，建议使用天数智芯官方适配的 AI 框架版本。
+为了发挥智铠100 的最佳性能，建议使用天数智芯官方适配的 corex 框架版本。
 
 ### 版本识别与安装
-官方适配的 Python 库通常会带有特定的标识（如 `ix` 或版本后缀）。
+官方适配的 Python 库通常会带有特定的标识（如 `corex` 或版本后缀）。
 *   **PyTorch**：需安装天数智芯适配版的 PyTorch。
 *   **安装建议**：强烈建议直接使用官方提供的 Docker 镜像进行开发，镜像中已预置了正确的驱动、CUDA 兼容库及 AI 框架。
 
 ```bash
 # 示例：检查环境中的适配包（具体名称以官方镜像为准）
-pip list | grep -E "torch|iluvatar|ix"
+pip list | grep corex
 ```
 
 ---
diff --git a/docs/compute/deploy/asr_model.md b/docs/compute/deploy/asr_model.md
index 3b7ad76436c7945ec513fe2c3b5f99b7d4dcd8cf..4abcb2e987f58df38003592b0bf69ced93061833 100644
--- a/docs/compute/deploy/asr_model.md
+++ b/docs/compute/deploy/asr_model.md
@@ -141,6 +141,60 @@ model = AutoModelForSpeechSeq2Seq.from_pretrained(
 )
 
 
+device = "cuda"
+model = model.to(device)
+
+# 加载音频 (强制重采样为 16kHz)
+audio, sr = librosa.load(audio_path, sr=16000)
+
+inputs = processor(audio, sampling_rate=sr, return_tensors="pt")
+
+input_features = inputs.input_features.to(device=device, dtype=torch.bfloat16)
+
+with torch.inference_mode():
+    generated_ids = model.generate(input_features, max_new_tokens=256)
+    text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
+
+print("识别结果:", text)
+```
+
+---
+### 三、天数智芯 (iluvatar) 部署指南 {#iluvatar-deploy}
+
+#### 1. 通用环境准备
+
+- **算力型号**: 智铠 100 (32GB)
+- **镜像选择**: `vLLM / 0.11.2 / Python 3.10 / IX-ML 4.4.0`
+
+#### 2. 模型部署实战
+
+##### 2.1 whisper-large-v3-turbo
+本示例演示如何在 **智铠 100** 环境完成语音识别。
+
+**运行推理代码**: 新建 Notebook 单元格运行。
+
+```python
+# 安装依赖
+!pip install transformers librosa 
+
+import torch
+import librosa
+from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor
+
+# 模型与音频路径
+model_name = "/mnt/moark-models/whisper-large-v3-turbo"
+audio_path = "/mnt/moark-models/github/index-tts/emo_sad.wav"
+
+processor = AutoProcessor.from_pretrained(model_name)
+
+model = AutoModelForSpeechSeq2Seq.from_pretrained(
+    model_name,
+    torch_dtype=torch.bfloat16,  # 这里指定了模型权重为 bfloat16
+    low_cpu_mem_usage=True,
+    use_safetensors=True
+)
+
+
 device = "cuda"
 model = model.to(device)
 
diff --git a/docs/compute/deploy/embedding_model.md b/docs/compute/deploy/embedding_model.md
index 6ec7a306ee0ce4f96fdd7274436bd0bf9fa6dcb0..c61f9f9185cced34ba2b59292f38b12023a44310 100644
--- a/docs/compute/deploy/embedding_model.md
+++ b/docs/compute/deploy/embedding_model.md
@@ -173,6 +173,61 @@ print(f"编码完成，生成 {len(texts)} 个向量")
 
 ---
 
+
+### 三、天数智芯 (iluvatar) 部署指南 {#iluvatar-deploy}
+
+#### 3.1 环境准备
+
+- **算力型号**：智铠 100 (32GB)
+- **镜像选择**：`vLLM / 0.11.2 / Python 3.10 / IX-ML 4.4.0`
+
+![镜像选择](/img/computility/iluvatar_image.jpg)
+
+#### 3.2 Embedding 推理示例
+
+```python
+!pip install transformers
+import torch
+import torch.nn.functional as F
+from torch import Tensor
+from transformers import AutoTokenizer, AutoModel
+
+#“句向量”绑定在最后一个有效 token上
+def last_token_pool(last_hidden_states: Tensor,
+                 attention_mask: Tensor) -> Tensor:
+    left_padding = (attention_mask[:, -1].sum() == attention_mask.shape[0])
+    if left_padding:
+        return last_hidden_states[:, -1]
+    else:
+        sequence_lengths = attention_mask.sum(dim=1) - 1
+        batch_size = last_hidden_states.shape[0]
+        return last_hidden_states[torch.arange(batch_size, device=last_hidden_states.device), sequence_lengths]
+
+
+model_name = "/mnt/moark-models/Qwen3-Embedding-8B"
+tokenizer = AutoTokenizer.from_pretrained(model_name, padding_side='left')
+model = AutoModel.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="cuda")
+
+texts = ["天空是蓝色的，天气晴朗。", "今天阳光灿烂，非常适合出游。"]
+
+batch_dict = tokenizer(
+    texts,
+    padding=True,
+    truncation=True,
+    max_length=8192,
+    return_tensors="pt"
+).to("cuda")
+
+batch_dict.to(model.device)
+outputs = model(**batch_dict)
+
+embeddings = last_token_pool(outputs.last_hidden_state, batch_dict['attention_mask'])
+print(embeddings)
+
+print(f"编码完成，生成 {len(texts)} 个向量")
+```
+
+---
 ## 三、vLLM 高性能部署
 
 vLLM 提供了对 Embedding 模型的优化支持，可直接生成 OpenAI 兼容的 API 接口，无需额外开发。
@@ -190,6 +245,8 @@ vLLM 会自动处理批处理、GPU 优化和 API 兼容性，开箱即用。
 
 ---
 
+
+
 ## 五、本地访问与服务验证
 
 请参考【[SSH 隧道配置指南](./ssh_tunnel.md)】建立安全连接。
diff --git a/docs/compute/deploy/img_model.md b/docs/compute/deploy/img_model.md
index 4de7cad7f4b8ce2a4876405a3ed69641034c12da..1e3ce89ee26b596d31984678324dd92f47c4c5e9 100644
--- a/docs/compute/deploy/img_model.md
+++ b/docs/compute/deploy/img_model.md
@@ -311,6 +311,83 @@ with torch.inference_mode():
 
 ---
 
+### 四、天数智芯 (iluvatar) 部署指南 {#iluvatar-deploy}
+
+本章节适用于 **智铠 100** 等沐曦系列算力卡。
+
+#### 1. 通用环境准备
+
+所有沐曦模型部署均需基于以下环境配置进行：
+
+- **算力型号**：智铠 100 (32GB)
+- **算力数量**：>=2
+- **镜像选择**: `vLLM / 0.11.2 / Python 3.10 / IX-ML 4.4.0`
+
+
+![镜像选择](/img/computility/iluvatar_image.jpg)
+
+**基础操作步骤：**
+1.  **进入工作台**：启动实例后，点击 **JupyterLab** 进入容器环境。
+
+    ![进入容器](/img/computility/iluvatar_Lab.jpg)  
+2.  **新建脚本**：点击 "Notebook" 图标，新建一个 `.ipynb` 文件。
+
+    ![新建Notebook](/img/computility/iluvatar_notebook.jpg)
+
+---
+
+#### 2. 模型部署实战
+
+请根据您需要的模型选择对应的实战案例代码。
+
+#### 2.1 Z-Image-Turbo
+本示例演示如何在 **智铠 100** 算力环境下，**加载平台内置模型**并生成高质量图像。
+
+**运行推理代码：**
+新建 Notebook 单元格运行
+
+```python
+import torch
+from diffusers import ZImagePipeline
+
+model_name = "/mnt/moark-models/Z-Image-Turbo"
+
+
+pipe = ZImagePipeline.from_pretrained(
+    model_name,
+    torch_dtype=torch.bfloat16,
+)
+pipe.to("cuda")
+
+prompt = '''A 9:16 vertical, realistic cyber-aesthetic future social profile card photo: A hand gently holds a vertically semi-transparent acrylic card, occupying the visual center of the picture. The card presents the personal homepage interface of the future social platform "MoArk", with a minimalist design and no redundant decorations. The edges of the card are rounded and soft, with a gradient neon halo of pink-purple and ice blue. The background is deep and blurry, further highlighting the crystal-clear texture of the card itself. The interface information seems to be micro-engraved, three-dimensional and clear, displayed in sequence:
+Avatar (suspended in the center, with a holographic surround effect)
+Username and the dynamic "Verified Member" badge at the top
+Name: MoArk (MoArk) Computing Power Experience Officer
+Followers: 2,777
+Following: 12,000
+Join Date: 11/7/2025
+Follow button (presenting a soft light touchable effect)
+A soft light and shadow is reflected at the point where the finger touches, creating an atmosphere that is both cinematic and immersive, like a scene from a near-future live-action game.'''
+
+image = pipe(
+    prompt=prompt,
+    height=1024,
+    width=1024,
+    num_inference_steps=9,  
+    guidance_scale=0.0,     
+    generator=torch.Generator("cuda").manual_seed(42),
+).images[0]
+
+image.save("image.png")
+print("图片生成完成，已保存为 image.png")
+```
+
+**查看结果与排查：**
+代码运行结束后，您可以在**左侧文件栏**找到 `image.png`，双击即可查看生成的图片。
+![图像结果](/img/computility/pytorch_img.png)
+
+---
+
 ## 四、本地访问与服务验证
 
 请参考【[SSH 隧道配置指南](./ssh_tunnel.md)】建立安全连接。
diff --git a/docs/compute/deploy/ocr_model.md b/docs/compute/deploy/ocr_model.md
index 6e3dbdd70ef552098c6bd9553764daeef824ee39..4cc203bbcb283617b65517866e9f4015a6c8ef2f 100644
--- a/docs/compute/deploy/ocr_model.md
+++ b/docs/compute/deploy/ocr_model.md
@@ -204,6 +204,86 @@ print(result)
 
 ---
 
+
+### 三、天数智芯 (iluvatar) 部署指南 {#iluvatar-deploy}
+
+#### 1. 通用环境准备
+
+- **算力型号**: 智铠 100 (32GB)
+- **镜像选择**: `vLLM / 0.11.2 / Python 3.10 / IX-ML 4.4.0`
+
+#### 2. 模型部署实战
+
+##### 2.1 PaddleOCR-VL-1.5
+本示例演示如何在 **智铠 100** 环境完成 OCR 识别。
+
+**运行推理代码**: 新建 Notebook 单元格运行。
+
+```python
+# 安装依赖
+!pip install "transformers>=5.0.0"
+
+import torch
+from PIL import Image
+from transformers import AutoProcessor, AutoModelForImageTextToText
+
+model_path = "/mnt/moark-models/PaddleOCR-VL-1.5"
+image_path = "/mnt/moark-models/L1_exam/ocr_demo.png"
+task = "ocr" # Options: 'ocr' | 'table' | 'chart' | 'formula' | 'spotting' | 'seal'
+
+image = Image.open(image_path).convert("RGB")
+orig_w, orig_h = image.size
+spotting_upscale_threshold = 1500
+
+# 通用图片预处理方式，目的是放大图片，针对spotting任务，需要确保图片内的文字清晰可见，以提高ocr的识别度，
+if task == "spotting" and orig_w < spotting_upscale_threshold and orig_h < spotting_upscale_threshold:
+    process_w, process_h = orig_w * 2, orig_h * 2
+    try:
+        resample_filter = Image.Resampling.LANCZOS
+    except AttributeError:
+        resample_filter = Image.LANCZOS
+    image = image.resize((process_w, process_h), resample_filter)
+
+max_pixels = 2048 * 28 * 28 if task == "spotting" else 1280 * 28 * 28
+
+DEVICE = "cuda"
+PROMPTS = {
+    "ocr": "OCR:",
+    "table": "Table Recognition:",
+    "formula": "Formula Recognition:",
+    "chart": "Chart Recognition:",
+    "spotting": "Spotting:",
+    "seal": "Seal Recognition:",
+}
+
+model = AutoModelForImageTextToText.from_pretrained(model_path, torch_dtype=torch.bfloat16).to(DEVICE).eval()
+processor = AutoProcessor.from_pretrained(model_path)
+
+messages = [
+    {
+        "role": "user",
+        "content": [
+            {"type": "image", "image": image},
+            {"type": "text", "text": PROMPTS[task]},
+        ]
+    }
+]
+inputs = processor.apply_chat_template(
+    messages,
+    add_generation_prompt=True,
+    tokenize=True,
+    return_dict=True,
+    return_tensors="pt",
+    images_kwargs={"size": {"shortest_edge": processor.image_processor.min_pixels, "longest_edge": max_pixels}},
+).to(model.device)
+
+outputs = model.generate(**inputs, max_new_tokens=512)
+result = processor.decode(outputs[0][inputs["input_ids"].shape[-1]:-1])
+print(result)
+```
+
+---
+
 ## 四、常见问题
 
 - **图片识别不完整**: 尝试提高图片分辨率或使用更清晰的扫描件。
diff --git a/docs/compute/deploy/speech_model.md b/docs/compute/deploy/speech_model.md
index b0f25154d6ae6662030f49dd1da273bc2dec4830..05eb9269ff110ccca2bcc68caca91cfc3f09273f 100644
--- a/docs/compute/deploy/speech_model.md
+++ b/docs/compute/deploy/speech_model.md
@@ -168,7 +168,7 @@ tts.infer(spk_audio_prompt='/mnt/moark-models/github/index-tts/emo_sad.wav', tex
 所有燧原模型部署均需基于以下环境配置进行：
 
 - **算力型号**：Enflame S60 (48GB)
-- **镜像选择**：`vLLM / 0.11.0 / Python 3.12 / ef 1.7.0.14`
+- **镜像选择**：`vLLM / 0.8.0 / Python 3.10 / ef 1.5.0.604`
 
 ![镜像选择](/img/computility/Enflame_flux-krea-dev_create.jpg)
 
@@ -233,6 +233,88 @@ tts.infer(spk_audio_prompt='/mnt/moark-models/github/index-tts/emo_sad.wav', tex
 
 ---
 
+### 三、天数智芯 (iluvatar) 部署指南 {#iluvatar-deploy}
+ 
+#### 1. 通用环境准备
+
+- **算力型号**：智铠 100 (32GB)
+- **镜像选择**：`vLLM / 0.11.2 / Python 3.10 / IX-ML 4.4.0`
+
+![镜像选择](/img/computility/iluvatar_image.jpg)
+
+**基础操作步骤：**
+1.  **进入工作台**：启动实例后，点击 **JupyterLab** 进入容器。
+
+    ![进入容器](/img/computility/iluvatar_Lab.jpg)
+2.  **新建脚本**：点击图标新建一个 `.ipynb` 文件。
+
+    ![新建Notebook](/img/computility/iluvatar_notebook.jpg)
+
+---
+
+#### 2. 模型部署实战
+
+请根据您需要的模型选择对应的实战案例代码。
+
+#### 2.1 IndexTTS-2
+本示例演示如何在 **智铠 100** 算力环境下，**加载平台内置模型**合成一段音频
+
+##### 运行推理代码
+**第一步安装相关的依赖**
+在notebook上新建单元格，运行以下代码
+```python
+#安装必要的依赖库
+!pip install bentoml \
+    accelerate==1.8.1 \
+    transformers==4.52.1 \
+    cn2an==0.5.22 \
+    ffmpeg-python==0.2.0 \
+    Cython==3.0.7 \
+    g2p-en==2.1.0 \
+    jieba==0.42.1 \
+    keras==2.9.0 \
+    numba \
+    numpy==1.26.2 \
+    pandas==2.1.3 \
+    matplotlib==3.8.2 \
+    opencv-python==4.9.0.80 \
+    vocos==0.1.0 \
+    omegaconf \
+    sentencepiece \
+    munch==4.0.0 \
+    librosa==0.10.2.post1 \
+    descript-audiotools==0.7.2 \
+    textstat==0.7.10 \
+    tokenizers==0.21.0 \
+    json5==0.10.0 \
+    pydub \
+    tqdm
+```
+![图像结果](/img/computility/speech_ix_1.png)
+
+**第二步加载模型**
+等待上述依赖安装完成后，在下方新起单元格，运行以下代码
+```python
+!cp -rf /mnt/moark-models/github/index-tts/indextts ./
+
+import os
+os.environ["USE_TF"] = "0"
+os.environ["USE_TORCH"] = "1"
+
+from indextts.infer_v2 import IndexTTS2
+#请不要修改cfg_path和model_dir的路径
+tts = IndexTTS2(cfg_path="/mnt/moark-models/github/index-tts/checkpoints/config.yaml", model_dir="/mnt/moark-models/IndexTTS-2", use_fp16=False, use_cuda_kernel=False, use_deepspeed=True)
+
+#如果需要生成不同的音色和音频内容，请替换spk_audio_prompt所需的参考音频以及text的输出内容。
+text = "快躲起来！是他要来了！他要来抓我们了！"
+tts.infer(spk_audio_prompt='/mnt/moark-models/github/index-tts/emo_sad.wav', text=text, output_path="gen.wav", emo_alpha=0.6, use_emo_text=True, use_random=False, verbose=True)
+```
+**查看结果与排查：**
+代码运行结束后，您可以在**左侧文件栏**找到 `gen.wav`，由于jupyter不支持播放音频文件，可下载至本地主机进行播放。
+![图像结果](/img/computility/speech_2.png)
+
+---
+
 ## 四、本地访问与服务验证
 
 请参考【[SSH 隧道配置指南](./ssh_tunnel.md)】建立安全连接。
diff --git a/docs/compute/deploy/text_model.md b/docs/compute/deploy/text_model.md
index 81471ddf739f92ba079269ac1fafca797cf3a014..0ce29fa0d43207a5f13e0609d250eb1c12fb5a90 100644
--- a/docs/compute/deploy/text_model.md
+++ b/docs/compute/deploy/text_model.md
@@ -53,61 +53,62 @@ Qwen3 系列提供从 0.6B 到 70B 不等的参数规模，原生支持 32K~128K
 2. 新建一个 `.ipynb` (Notebook) 文件。
    ![新建Notebook](/img/computility/pytorch0.png)
 3. 安装依赖：在第一个单元格运行以下命令。安装完成后，请务必重启内核 (Restart Kernel) 以使环境生效。
-   ```bash
-   pip install accelerate
-   ```
-   ![重启内核](/img/computility/pytorch2.png)
-4. 运行推理代码：沐曦环境支持直接加载挂载盘中的内置模型，无需下载。
-
-   ```python
-   from transformers import AutoModelForCausalLM, AutoTokenizer
-   import torch
-
-   # 指定内置模型库路径 (无需下载)
-   model_name = "/mnt/moark-models/Qwen3-0.6B"
-
-   tokenizer = AutoTokenizer.from_pretrained(model_name)
-   model = AutoModelForCausalLM.from_pretrained(
-       model_name,
-       torch_dtype="auto",
-       device_map="auto"
-   )
-
-   prompt = "Give me a short introduction to large language model."
-   messages = [{"role": "user", "content": prompt}]
-
-   text = tokenizer.apply_chat_template(
-       messages,
-       tokenize=False,
-       add_generation_prompt=True,
-       enable_thinking=True
-   )
-
-   model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
 
-   generated_ids = model.generate(
-       **model_inputs,
-       max_new_tokens=32768
-   )
-
-   output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
-   try:
-       index = len(output_ids) - output_ids[::-1].index(151668)  # </think>
-   except ValueError:
-       index = 0
+```bash
+pip install accelerate
+```
+![重启内核](/img/computility/pytorch2.png)
+4. 运行推理代码：沐曦环境支持直接加载挂载盘中的内置模型，无需下载。
 
-   thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")
-   content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n")
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+
+# 指定内置模型库路径 (无需下载)
+model_name = "/mnt/moark-models/Qwen3-0.6B"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+      model_name,
+      torch_dtype="auto",
+      device_map="auto"
+)
+
+prompt = "Give me a short introduction to large language model."
+messages = [{"role": "user", "content": prompt}]
+
+text = tokenizer.apply_chat_template(
+      messages,
+      tokenize=False,
+      add_generation_prompt=True,
+      enable_thinking=True
+)
+
+model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
+
+generated_ids = model.generate(
+      **model_inputs,
+      max_new_tokens=32768
+)
+
+output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
+try:
+      index = len(output_ids) - output_ids[::-1].index(151668)  # </think>
+except ValueError:
+      index = 0
+
+thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")
+content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n")
+
+print("=== Thinking ===")
+print(thinking_content)
+print("\n=== Response ===")
+print(content)
+```
 
-   print("=== Thinking ===")
-   print(thinking_content)
-   print("\n=== Response ===")
-   print(content)
-   ```
+执行结果示例：
 
-   执行结果示例：
-   
-   ![执行结果](/img/computility/pytorch3.png)
+![执行结果](/img/computility/pytorch3.png)
 
 #### 2. vLLM 高性能推理
 
@@ -122,14 +123,19 @@ Qwen3 系列提供从 0.6B 到 70B 不等的参数规模，原生支持 32K~128K
 1. 通过 JupyterLab 打开终端 (Terminal)。
 2. 执行以下命令启动兼容 OpenAI 接口的服务：
 
-   ```bash
-   # 使用内置模型路径，指定端口 8188
-   vllm serve /mnt/moark-models/Qwen3-0.6B --port 8188
-   ```
-   ![vLLM启动命令](/img/computility/vllm2.png)
+```bash
+# 使用内置模型路径，指定端口 8188
+vllm serve /mnt/moark-models/Qwen3-0.6B --port 8188
+```
+
+:::info
+若需要指定model的id，可通过--served-model-name Qwen3-0.6B重新调整
+:::
+
+![vLLM启动命令](/img/computility/vllm2.png)
 
 3. 当看到 `Uvicorn running on http://0.0.0.0:8188` 字样时，服务启动成功。
-   ![vLLM运行成功](/img/computility/vllm3.png)
+![vLLM运行成功](/img/computility/vllm3.png)
 
 #### 3. SGLang 结构化推理
 
@@ -142,12 +148,12 @@ Qwen3 系列提供从 0.6B 到 70B 不等的参数规模，原生支持 32K~128K
 1. 通过 JupyterLab 打开终端 (Terminal)。
 2. 执行以下命令启动服务：
 
-   ```bash
-   # SGLang 启动命令
-   python -m sglang.launch_server \
-     --model-path /mnt/moark-models/Qwen3-0.6B \
-     --port 8188
-   ```
+```bash
+# SGLang 启动命令
+python -m sglang.launch_server \
+   --model-path /mnt/moark-models/Qwen3-0.6B \
+   --port 8188
+```
    ![SGLang启动命令](/img/computility/sglang2.png)
 
 3. 服务启动成功界面：
@@ -175,37 +181,37 @@ Qwen3 系列提供从 0.6B 到 70B 不等的参数规模，原生支持 32K~128K
    ![进入容器](/img/computility/Enflame_flux-krea-dev_notebook.jpg)
 3. 代码迁移关键点：必须在代码最前方引入 `torch_gcu` 库。
 
-   ```python
-   import torch
-   import torch_gcu
-   from torch_gcu import transfer_to_gcu
-   from transformers import AutoModelForCausalLM, AutoTokenizer
+```python
+import torch
+import torch_gcu
+from torch_gcu import transfer_to_gcu
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-   model_name = "/mnt/moark-models/Qwen3-0.6B"
+model_name = "/mnt/moark-models/Qwen3-0.6B"
 
-   tokenizer = AutoTokenizer.from_pretrained(model_name)
-   model = AutoModelForCausalLM.from_pretrained(
-       model_name,
-       torch_dtype="auto",
-       device_map="auto"
-   )
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+      model_name,
+      torch_dtype="auto",
+      device_map="auto"
+)
 
-   prompt = "Explain quantum computing in simple terms."
-   messages = [{"role": "user", "content": prompt}]
+prompt = "Explain quantum computing in simple terms."
+messages = [{"role": "user", "content": prompt}]
 
-   text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-   model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
+text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
 
-   generated_ids = model.generate(**model_inputs, max_new_tokens=512)
-   output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
-   response = tokenizer.decode(output_ids, skip_special_tokens=True)
-   print(response)
-   ```
+generated_ids = model.generate(**model_inputs, max_new_tokens=512)
+output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
+response = tokenizer.decode(output_ids, skip_special_tokens=True)
+print(response)
+```
 
 #### 2. vLLM 高性能推理
 
 #### 2.1 环境准备
-- 镜像选择：`Ubuntu / 22.04 / Python 3.13 / ef 1.5.0.604`
+- 镜像选择：`vLLM / 0.11.0 / Python 3.12 / ef 1.7.0.14`
   （注：该官方镜像已内置 vLLM 适配环境）
 
 ![镜像选择](/img/computility/Enflame_flux-krea-dev_create.jpg)
@@ -217,6 +223,10 @@ Qwen3 系列提供从 0.6B 到 70B 不等的参数规模，原生支持 32K~128K
 vllm serve /mnt/moark-models/Qwen3-0.6B --port 8188
 ```
 
+:::info
+若需要指定model的id，可通过--served-model-name Qwen3-0.6B重新调整
+:::
+
 ![vLLM启动命令](/img/computility/Enflame_Qwen3-0.6B_vllm_00.jpg)
 
 ![进入容器](/img/computility/Enflame_Qwen3-0.6B_vllm_01.jpg)
@@ -229,6 +239,67 @@ vllm serve /mnt/moark-models/Qwen3-0.6B --port 8188
 
 ---
 
+### 三、天数智芯 (iluvatar) 部署指南 {#iluvatar-deploy}
+
+#### 1 Hugging Face Transformers
+
+#### 1.1 环境准备
+- **算力型号**：智铠 100 (32GB)
+- **镜像选择**：`vLLM / 0.11.2 / Python 3.10 / IX-ML 4.4.0`
+
+![镜像选择](/img/computility/iluvatar_image.jpg)
+
+#### 1.2 部署步骤
+1. 启动实例后，进入 JupyterLab。
+   ![进入容器](/img/computility/iluvatar_Lab.jpg)
+2. 新建一个 `.ipynb` (Notebook) 文件。
+   ![进入容器](/img/computility/iluvatar_notebook.jpg)
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_name = "/mnt/moark-models/Qwen3-0.6B"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+      model_name,
+      torch_dtype="auto",
+      device_map="auto"
+)
+
+prompt = "Explain quantum computing in simple terms."
+messages = [{"role": "user", "content": prompt}]
+
+text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
+
+generated_ids = model.generate(**model_inputs, max_new_tokens=512)
+output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
+response = tokenizer.decode(output_ids, skip_special_tokens=True)
+print(response)
+```
+
+#### 2. vLLM 高性能推理
+
+#### 2.1 环境准备
+- **算力型号**：智铠 100 (32GB)
+- **镜像选择**：`vLLM / 0.11.2 / Python 3.10 / IX-ML 4.4.0`
+
+![镜像选择](/img/computility/iluvatar_image.jpg)
+
+#### 2.2 启动服务
+在终端中执行以下命令：
+
+```bash
+vllm serve /mnt/moark-models/Qwen3-0.6B --port 8188
+```
+:::info
+若需要指定model的id，可通过--served-model-name Qwen3-0.6B重新调整
+:::
+
+---
+
 ## 四、本地访问与服务验证
 
 请参考【[SSH 隧道配置指南](./ssh_tunnel.md)】建立安全连接。
diff --git a/static/img/computility/iluvatar_Lab.jpg b/static/img/computility/iluvatar_Lab.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..cbef4ae85d7b244d60c450afc42d470f529511d3
Binary files /dev/null and b/static/img/computility/iluvatar_Lab.jpg differ
diff --git a/static/img/computility/iluvatar_image.jpg b/static/img/computility/iluvatar_image.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..51e386fececcf834cde6c7497aee5a3f219b7398
Binary files /dev/null and b/static/img/computility/iluvatar_image.jpg differ
diff --git a/static/img/computility/iluvatar_notebook.jpg b/static/img/computility/iluvatar_notebook.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..948e0c658f46c2cf3c1ae147cba9b659cda98892
Binary files /dev/null and b/static/img/computility/iluvatar_notebook.jpg differ
diff --git a/static/img/computility/speech_ix_1.png b/static/img/computility/speech_ix_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..ab4d66d275b5cb03ee025eefd89581549b08e161
Binary files /dev/null and b/static/img/computility/speech_ix_1.png differ