Thud Audio Action Model Responds in 200ms, 10x Faster Than Gemini Live

Developers demo Thud, billed as the first Audio Action Model that acts directly on speech without transcription. Running fully locally with 200ms latency, it is claimed to be 10x faster and 20x cheaper than Gemini Live.

2026-10-14 ~ 2026-10-15 · 2 related posts

• Thud Audio Action Model: Voice-Controls Smart Glasses in 200ms, Fully Local — AnhPhuNguyen1 · 2026-10-14
• Thud Audio Action Model Claims 10x Faster, 20x Cheaper Than Gemini Live — AnhPhuNguyen1 · 2026-10-15