SmolLM2-360M · Flash Attention · RoPE · KV-Cache
NeuralAI
A production-grade ChatGPT-style LLM built from scratch in PyTorch. Memory, tools, streaming, voice — all yours.
💬
Chat
Streaming SSE responses with memory & conversation history
📁
NeuralDrive
Upload & analyze files. Cloud file system built in.
💻
Terminal
Real AI-powered shell — Bash, Python, Node.js
🎤
Live Voice
Speech-to-speech conversations with persona voices