Nexa AI

nexa.ai
On the map Visit site

Run any AI model on any device, fast, private, and offline.

Description

Nexa AI Is Now Part of Qualcomm AI Hub. · The platform for on-device AI, with optimized open source and licensed models, or bring your own. Validate performance on real Qualcomm devices.

Nexa AI – это комплексный хаб AI-моделей и фреймворк вывода для устройств, созданный для того, чтобы разработчики и предприятия могли локально развертывать и запускать более 700 квантизированных AI-моделей на различных периферийных устройствах. Он поддерживает несколько модальностей, включая текст, зрение и аудио, и оптимизирован для CPU, GPU и NPU на таких платформах, как Windows, macOS, Linux, Android и iOS. Nexa AI подчеркивает конфиденциальность, низкую задержку и экономическую эффективность, устраняя зависимость от облачных сервисов, предоставляя при этом такие инструменты, как Nexa SDK для бесшовного развертывания в одну строку и аппаратного ускорения. Платформа также способствует созданию совместного сообщества для обмена моделями и поддержки разработки, делая AI на устройствах практичным и масштабируемым.

Features

Absolute Privacy
Predictable Cost Model
Offline Reliability
Broad Hardware Compatibility
Any Model, Any Device Deployment
NPU & GPU Acceleration
Model Optimization & Compression
Cross-Platform Development
Unlimited Local File Context
Agentic RAG with Vision

FAQ

Nexa AI is a platform and SDK that allows running AI models (including large language models, vision-language models, speech recognition, and text-to-speech) locally on CPUs, GPUs, and NPUs across multiple hardware and operating systems, ensuring privacy and low-latency performance without Internet dependency.

Nexa AI supports developers and enterprises who want to deploy AI models offline with high precision and speed on resource-constrained devices, without requiring extensive hardware resources or cloud connectivity.

It supports state-of-the-art models from various providers, including DeepSeek, Llama, Gemma, Qwen, and Nexa's own models (Octopus, OmniVLM, OmniAudio), covering tasks such as text generation, image generation, speech transcription, and function calling.

Nexa AI works on desktop, mobile, automotive, and IoT devices, compatible with Qualcomm, Intel, AMD chipsets, and supports Windows, macOS, Linux, Android, and iOS systems.

Developers can use the Nexa AI SDK (available in Python, Go, and C++) to integrate AI models locally. Installation and model setup are simplified, including features like GPU acceleration, resource allocation, and model selection without coding complexity.

Nexa AI runs models offline, keeping data private on-device. It offers significant speed and energy efficiency improvements (e.g., 9x faster multimodal tasks, 35x faster function calls, and model compression technologies like NexaQuant that maintain accuracy while boosting speed).

Typical use cases include building private AI chatbots, voice assistants with on-device automatic speech recognition and speech synthesis, multimodal AI applications, and enterprise-grade AI deployment with secure scaling.

Specs

Type Agent
SectionInfrastructure & MLOps
Pricing free (от $0/mo)
Platform Cross-platform
Systems windows, android, macos, linux, ios, web
Hostingself-hosted
Who forIndividual
Site languageen
GitHubNexaAI/nexa-sdk
Rating4.50 (0 reviews)
Views4 068

Integrations

GitHub slack Google IBM Granite Qualcomm AMD

Platforms

web macos ios android windows linux

Tags

local inference data privacy solutions

Hashtags

agent

Social

GitHub Discord LinkedIn

Source code

NexaAI/nexa-sdk

Found in sources

aiagentslist moge

Similar in «Infrastructure & MLOps»

AWS
Amazon’s cloud: Bedrock, SageMaker and the infrastructure under most AI services.
HPE GreenLake
Комплексная edge-to-cloud платформа, предоставляющая гибкую ИТ-инфраструктуру и управление как услуга для гибридных сред.
Langfuse
LLM engineering platform for observability and metrics
Needle
Boost productivity: Automate workflows and find answers with AI.
CrewAI
Automate complex tasks with AI agents working together.
Vessium
Automate complex business tasks by describing them in plain words.

Submit a site to the catalog

Just send the link — we will work out the rest.

We will review what you send and add it to the catalog if it fits.