OmAI Secures Millions in Funding, Open-Sources Edge-Native Multimodal Model VLX-Seek1.5
量子位 · wechat · 2026-08-06
Embodied AI company OmAI announced hundreds of millions in funding and open-sourced its 'edge-native' multimodal model, VLX-Seek1.5 (3B and 10B versions).
- Edge-Native Architecture: Unlike traditional models compressed for edge deployment, VLX is designed from the ground up with edge compute, latency, and power constraints as foundational principles.
- Streaming Multimodal: Moves beyond batch-processing single frames to continuously process video streams, forming a 'Flow-Seek-Go' closed loop for real-time understanding.
- Performance: In drone-view benchmarks (RefDrone), the 3B model significantly outperforms similarly sized models like Nvidia's Locate Anything, while drastically reducing object hallucination rates (18 vs. 71.3).
- Ecosystem: The company aims to be the infrastructure for physical AI, having deployed the OmAgent platform and Homer AI, a wearable visual hub for the visually impaired.
More from Venture
- Creator AI SaaS Reclip Acquired for $30K — marclou · 2026-08-06
- PowerBot's Multi-Sport AI Coach Robot Raises Over $4M on Kickstarter — 创业邦 · 2026-08-06
- Amazon's Zoox to Launch Paid Fully Driverless Robotaxi Rides in Las Vegas — Polymarket · 2026-08-06
- Ropedia raises tens of millions in funding to build real-world experience infrastructure for Physical AI — 机器之心 · 2026-08-06
- Easy-to-Build AI Products Have the Thinnest Moats: A Warning for Indie Devs — sven_ai · 2026-08-06
- Monolith Seeks $500M for New Fund Targeting Chinese AI Startups — pstAsiatech · 2026-08-06