Modern technology to bring the most advanced virtual companion straight to your browser.
Connect to 20+ leading language model providers including OpenAI, Claude, DeepSeek, Gemini, and more. Haruka adapts to your preferred model and maintains rich, context-aware conversations that feel genuinely natural. Switch providers anytime without losing your conversation style.
Natural voice interaction using advanced speech recognition and text-to-speech technologies. Just talk: Haruka listens and responds in real time.
Capable of responding and understanding popular video games in real-time. Haruka can react and comment alongside you while you play.
Full support for high-speed, secure, and completely private local AI processing. Haruka runs AI inference directly on your GPU, zero data leaves your device.
Highly expressive 3D models moving dynamically in response to touch and conversational voice. Every interaction triggers real, lifelike animations.
Connectable to 20+ large language model providers including OpenAI, Claude, and DeepSeek. One companion, unlimited intelligence options.
Native English and Japanese support built in. Practice your Japanese naturally with Haruka, she's patient, encouraging, and always happy to help.
Access your companion anywhere with a responsive web interface and native-like app installation. Works offline as a PWA.
An extensible component architecture for custom skills, tools, and integrations. Build and share your own companion enhancements with the community.
Full control over your setup. Deploy Haruka on your own infrastructure: your data, your servers, your rules. No vendor lock-in, ever.