AutoArk/EVA-OS

EVA OS — A real-time multimodal AIOS for next-generation hardware, enabling your devices being “alive” and as intelligent as a real brain.

What it solves

EVA Platform is an AIOS designed to AI application and smart hardware developers to build real-time multimodal AI applications. It removes the complexity of building audio-visual links, adapting to various model interfaces, and managing cloud resources, allowing developers to focus on the product experience.

How it works

The platform combines agentic development tools, an on-device runtime, and a model gateway to streamline the creation of AI applications:

  • AI Native Development: Uses "Agent Skills" to allow developers to describe product requirements in natural language, enabling AI agents to handle project initialization, code development, and debugging.
  • EVA SDK: Provides TypeScript and Python SDKs for apps, IoT, and embedded devices to implement full-duplex audio-visual interaction, emotion recognition, and device command execution.
  • Model Gateway: A unified interface for ASR, TTS, and LLM providers, allowing developers to switch models via a single parameter without rewriting business logic.
  • Edge AI: Integrates proprietary on-device ASR and TTS models to reduce latency and cloud dependency.
  • EVA CLI: A command-line tool for managing projects, API keys, and cloud resources, which can also be operated by AI agents.

Who it’s for

Developers building AI companion hardware (toys, robots), voice-interactive devices (smart speakers, wearables), real-time AI apps (language practice, customer service), and IoT/embodied intelligence projects.

Highlights

  • Vibe Coding: Natural language driven development using Agent Skills.
  • Full-Duplex Interaction: Low-latency real-time communication with support for natural interruptions.
  • SOTA Performance: Proprietary ASR model ranked first on the Hugging Face Open ASR Leaderboard for full-size models.
  • Hardware Agnostic: Support for TypeScript, Python, and upcoming support for Flutter, ESP32, and other embedded platforms.

Related

  • Project
    espressif/esp-brookesia

    A full-stack development platform for AIoT and HMI products that decouples hardware from software to enable scalable app deployment and AI agent integration on ESP devices.

  • Project
    elizaOS/eliza

    An open-source TypeScript framework and product stack for creating autonomous AI agents with a plugin-based architecture and support for local on-device inference.

  • Project
    agiresearch/AIOS

    AIOS is an open‑source “operating system” for LLM‑based agents. The kernel (this repo) manages LLM instances, memory, storage, and tool sandboxing, exposing a syscall‑style API via the Cerebrum SDK. It supports many model back‑ends (Ollama, vLLM, HuggingFace, Deepseek), integrates external tools, and offers several deployment modes—from local single‑machine setups to remote, personal, and virtualized kernels. Installation is via a Python 3.10/3.11 environment, configuration is done in `config.yaml`, and agents can be run through a web UI or a natural‑language terminal. The project is research‑grade (multiple conference papers) and actively maintained.

  • Project
    duixcom/Duix-Mobile

    An open-source SDK for deploying real-time interactive AI avatars on mobile and embedded devices with low latency and on-device execution.

  • Project
    dappros/ethora

    A chat and AI SDK platform that allows developers to integrate real-time messaging, LLM-powered bots, and RAG pipelines into web and mobile applications.