Radar / AI robotics / VAETKI Vision 3B
VAETKI Vision 3B: NC AI's ultra-light VLM for robots and field equipment
A 3B vision-language model built to run directly on robots and industrial devices with limited compute, reading video in 4-second, 8-frame units, with a welding-specialized variant that lifts weld classification accuracy from 41.1 to 80.9 in NC AI's tests.
Why it matters
It pushes robot vision onto the device itself instead of the cloud, and NC AI positions it as the perception core for future vision-language-action models.
What you could build with it
A factory-automation team could mount VAETKI Vision 3B on inspection robots and edge cameras to flag weld defects and equipment states on site, without streaming video to the cloud, and feed its scene understanding into a robot controller for simple pick-and-check tasks.
Does it hold up?
Too early to judge: announced 3 Nov 2026 with vendor benchmark scores (63.8 average on BLINK, ERIQ and ERQA) but no public weights, demo, or independent deployment yet.
Built with VAETKI Vision 3B
- NC AI company profile: VAETKI engine and physical-AI expansionarticle · Profiles NC AI as NCSOFT's AI affiliate and VAETKI as its multimodal engine for industry, with a March 2026 world foundation model demo for robot intelligence behind this vision release.
- NC AI's VAETKI beats OpenAI and Meta models on Korean benchmarksarticle · Earlier VAETKI 100B coverage showing the model family's Korean-language and instruction-following benchmark claims, giving context for the new 3B vision variant.
Learn more
First spotted on article: source.
More AI robotics
RoboParty RP1: a full-stack open-source humanoid robot for embodied AI researchRoboParty's 1.3-metre humanoid, launched 16 October and back in coverage on 2 November, comes in 24-DoF standard and…robotics · JEV 0.62FineART: Scale AI and Hugging Face's open robotics dataset and plan-before-act VLA modelFineART is the largest open robot dataset with subtask labels: 40,543 episodes, 1,718 hours of footage, 151 tasks and…robotics · JEV 0.59Bear Robotics KINO-2: demonstration-to-skill robot-learning platformBear Robotics launched KINO-2 on Oct 21 at GTC Berlin: a robot-learning platform that turns human task demonstrations…robotics · JEV 0.59Dream-1: robot foundation model that controls unseen bimanual robots zero-shot, at $3/hourDream Labs came out of stealth on 19 October with Dream-1, a robot foundation model offered through a public Dream API…robotics · JEV 0.52AGIBOT Innovation Day: GO-3 Preview, WITA-Omni, G2 Max, Link-U OS 1.0 and the Worken productivity metricShanghai embodied-AI firm AGIBOT unveiled its GO-3 Preview embodied foundation model, WITA-Omni interaction model…robotics · JEV 0.43
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.