A developer showcases a fully local voice chatbot running Qwen3.5-397B, Whisper-small, and Orpheus TTS with real-time streaming and interruption recovery. The chatbot, named Athena, engages in deep philosophical discussions about consciousness and self-preservation.
I've been working on this after hours for a few months continuously improving it. Now at a point where the chatbot is close to real-time (thanks to SSE streaming) and also interruptible while preserving context of what was last said. 100% local and powered by Qwen3.5-397B (Unsloth's UD-Q3\_K\_XL), Whisper-small STT, and Orpheus Q4\_K\_XL TTS with a custom SNAC decoder on ONNX. VRAM usage holds at 21.3 GB or less leaving decent headroom for compute graphs on a 24 GB GPU. System RAM MoE experts for Qwen occupy about \~150 GB. This is running with bf16 KV cache (Qwen3.5 spazzes out with Q8 KV), at 131,072 tokens. Enough for hours of conversation. GitHub code coming soon - should be able to upload this evening after I'm done with the honey-do list.