> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nexa.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Quickstart
> Get started with NexaSDK on MacOS, Windows, and Linux in 1 minute.
## **Installation & Running Your First Model**
### **MacOS Installation**
Download the appropriate installer for your Mac:
* [arm64 with Apple Neural Engine support](https://nexa-model-hub-bucket.s3.us-west-1.amazonaws.com/public/nexa_sdk/downloads/nexa-cli_macos_arm64.pkg)
* [x86\_64](https://nexa-model-hub-bucket.s3.us-west-1.amazonaws.com/public/nexa_sdk/downloads/nexa-cli_macos_x86_64.pkg)
Run the downloaded `.pkg` file and follow the installation wizard.
### **Running Your First Model**
MacOS supports both MLX (Apple Silicon optimized) and GGUF models.
**Language Model (LLM)**
```bash MacOS theme={"dark"}
nexa infer NexaAI/Qwen3-4B-4bit-MLX
```
**Multimodal Model**
```bash MacOS theme={"dark"}
nexa infer NexaAI/gemma-3n-E4B-it-4bit-MLX
```
Many MLX models in the Hugging Face mlx-community have quality issues and may not run locally. We recommend using models from [our collection](https://huggingface.co/NexaAI/collections) for best results.
**Language Model (LLM)**
```bash MacOS theme={"dark"}
nexa infer NexaAI/Qwen3-0.6B
```
**Multimodal Model**
```bash MacOS theme={"dark"}
nexa infer NexaAI/Qwen2.5-Omni-3B-GGUF
```
To try other GGUF models, visit Hugging Face, copy the path of any compatible GGUF model (e.g., `unsloth/Qwen2.5-VL-3B-Instruct-GGUF`), and replace the model path in the command above.
### **Windows x64 Installation**
Download the installer:
* [x86\_64 with Intel NPU support](https://nexa-model-hub-bucket.s3.us-west-1.amazonaws.com/public/nexa_sdk/downloads/nexa-cli_windows_x86_64.exe)
Run the downloaded `.exe` file and follow the installation wizard.
### **Running Your First Model**
Currently, we support LLM and Multimodal models. More model type support is coming soon!
**Language Model (LLM)**
```bash Windows theme={"dark"}
nexa infer NexaAI/Qwen3-0.6B
```
**Multimodal Model**
```bash Windows theme={"dark"}
nexa infer NexaAI/Qwen2.5-Omni-3B-GGUF
```
To try other GGUF models, visit Hugging Face, copy the path of any compatible GGUF model (e.g., `unsloth/Qwen2.5-VL-3B-Instruct-GGUF`), and replace the model path in the command above.
Currently, **LLM** (Large Language Model) and **VLM** (Vision Language Model) are in the testing scope. More modalities are coming soon!
### **Windows ARM64 Installation**
Download the installer:
* [arm64 with Qualcomm NPU support](https://nexa-model-hub-bucket.s3.us-west-1.amazonaws.com/public/nexa_sdk/downloads/nexa-cli_windows_arm64.exe)
Run the downloaded `.exe` file and follow the installation wizard.
### **Running Your First Model**
Currently, we support LLM and Multimodal models. More model type support is coming soon!
**Language Model (LLM)**
```bash Windows theme={"dark"}
nexa infer NexaAI/Qwen3-0.6B
```
**Multimodal Model**
```bash Windows theme={"dark"}
nexa infer NexaAI/Qwen2.5-Omni-3B-GGUF
```
To try other GGUF models, visit Hugging Face, copy the path of any compatible GGUF model (e.g., `unsloth/Qwen2.5-VL-3B-Instruct-GGUF`), and replace the model path in the command above.
Currently, **LLM** (Large Language Model) and **VLM** (Vision Language Model) are in the testing scope. More modalities are coming soon!
### **NPU Acceleration (Snapdragon X Elite)**
**Hardware requirement:** The following NPU-accelerated model currently runs only on Qualcomm Snapdragon X Elite laptops.
If you have a Snapdragon X Elite PC, you can run the flagship OmniNeural-4B model with NPU acceleration:
**OmniNeural-4B (Multimodal NPU Model)**
```bash Windows theme={"dark"}
nexa infer omni-neural
```
**Voice Input Mode:** Once running, record your voice directly in terminal:
```bash Windows theme={"dark"}
> /mic
```
Press CTRL + C to stop recording, then hit enter to send.
**File Input:** Drag image/audio files into the command line:
```
> describe this image '/path/to/image.jpg' '/path/to/audio.wav'
```
For detailed NPU setup instructions and advanced features, see the [NPU Guide](/en/nexa-sdk-go/npu).
### **Linux Installation**
Run the following command to download and install:
```bash Linux theme={"dark"}
curl -fsSL /path/to/install.sh -o install.sh && chmod +x install.sh && ./install.sh
```
### **Running Your First Model**
Currently, we support LLM and Multimodal models. More model type support is coming soon!
**Language Model (LLM)**
```bash Linux theme={"dark"}
nexa infer NexaAI/Qwen3-0.6B
```
**Multimodal Model**
```bash Linux theme={"dark"}
nexa infer NexaAI/Qwen2.5-Omni-3B-GGUF
```
To try other GGUF models, visit Hugging Face, copy the path of any compatible GGUF model (e.g., `unsloth/Qwen2.5-VL-3B-Instruct-GGUF`), and replace the model path in the command above.
We currently support **LLM** (Large Language Model) and **VLM** (Vision Language Model). More modalities are coming soon!
## **Explore CLI Commands**
To see a list of all available CLI commands, run:
```bash MacOS theme={"dark"}
nexa -h
```
```bash Windows theme={"dark"}
nexa -h
```
```bash Linux theme={"dark"}
nexa -h
```