New Team Noiz Unveils Real-Time Interactive Multimodal Model at 1% of Mainstream Inference Cost
新智元 · wechat · 2026-08-07
New team Noiz has unveiled a real-time interactive multimodal model with inference costs at just 1% of current mainstream audio-video models. By avoiding language as a universal translator, building event-centric representations, and layering an interaction layer atop base models, they achieve low-latency continuous responses. Their product noiz.ai has attracted 2 million creators, with revenue growing at high double-digit month-over-month and annualized revenue reaching millions of dollars. Founder Vega Chen previously led ASR model deployment at Meta FAIR, and co-founder Chen Qian has prior startup experience with nearly 10 million paying users.
More from Companies & People
- Clean Code Author Open-Sources swarm-forge for Coordinating AI Agents — unclebob · 2026-08-07
- ComfyUI Hits 123K Stars: Anonymous Dev Challenges Adobe's Creative Empire — msharmas · 2026-08-07
- SemiAnalysis: DeepMind No Longer a Frontier Lab, Talent Exodus and Compute Misallocation — firstadopter · 2026-08-07
- Grok Build V1.0 Lacks New Model, Only UI Fixes, Lags Behind Rivals — eyishazyer · 2026-08-07
- Anthropic Exec Predicts AI Agent Traffic Will Be 1000x Human Traffic in 5 Years — threepointone · 2026-08-07
- Corporate AI Insight: Shared Working Memory is the Core Asset — sarahdrinkwater · 2026-08-07