This release is a pre-release and may not be stable for production use.
nexaai
Run next-generation LLMs and VLMs locally, natively, and at top speed on any hardware.
nexaai is the official Python SDK for NexaSDK/NexaML:
An NPU‑first cross‑platform inference engine that powers LLMs, VLMs, embeddings, audio, and vision models across NPU, GPU, and CPU—on desktop, mobile, automotive, and IoT.
Features
- 🚀 NPU‑First Performance — Accelerated AI with automatic device selection and backend optimizations.
- 🔮 Day‑0 Architecture Support — Run new LLMs, VLMs, ASR, and CV models immediately (GGUF, MLX, .nexa, and more).
- 🧩 True Multimodality — Seamless pipelines for text, vision, and audio.
- 🤖 OpenAI-Compatible Server — Supports serving, chat, and function calling.
- 🏆 Cross‑Platform — macOS (Apple Silicon), Windows (x64, ARM64), Linux, mobile, embedded.
What is it?
nexaai offers a clean Python API over NexaML’s runtime. Build modern AI applications—chatbots, copilots, vision tools—that run fully on-device with maximal hardware utilization. Designed for rapid adoption of new models and formats, especially with NPU acceleration.
Installation & Quickstart
Please refer to our official docs for up-to-date install steps, environment notes, and code examples:
- Overview: NexaAI Python SDK Overview
- Quickstart: Quickstart Guide
The docs include supported Python versions, backend requirements (NPU, MLX, GPU, CPU), and ready-to-run examples for LLMs, VLMs, embeddings, and audio.
Links
- 📚 Docs: https://docs.nexa.ai/nexa-sdk-python/overview
- 🐞 Issues/Discussions: NexaAI Issues
- 💬 Community:
2 maintainers
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
File details
Details for the file nexaai-1.0.37rc10.tar.gz.
File metadata
- Download URL: nexaai-1.0.37rc10.tar.gz
- Upload date:
- Size: 61.4 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.10.19
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
ec0266fb574b9d338fc885aa90bd400f94fd10cf8c851963e2c96729da84690d
|
|
| MD5 |
0ebb205519e83c909a00b4d40aadf065
|
|
| BLAKE2b-256 |
e2c553615c413ef362a1a02ead4db6f015c620d3951dc58a00a8ae513a6507f3
|
