Skip to content

RobotsNAV·01Robots VideosNAV·02Videos ComponentsNAV·03Components PapersNAV·04Papers BlogNAV·05Blog Tech CrowdfundingNAV·06Tech Crowdfunding Open SourceNAV·07Open Source World ModelsNAV·08World Models LeaderboardNAV·09Leaderboard Industry ChainNAV·10Industry Chain InvestNAV·11Invest Creative WorkshopNAV·12Creative Workshop

physical-ai,skills,applications,sota,demos,llm,harness,image,video,world3d,design,office,ai-games,about

[●●●]FETCHING

RobotWorld

Robotics knowledge and video hub · © 2026

AgientryAGI intelligence hub

@McQueenFuFollow Tags About DMCA1003 demos340 papers873 components

← Back to Videos

Reka EdgeQ: An On-Device VLM Running Natively on the Snapdragon Hexagon NPU

Loading video

Loading video

Reka EdgeQ is an optimized on-device VLM running natively on the Qualcomm Snapdragon 8 Elite Hexagon NPU: 0.73s time to first token on images, +34 points over Gemma 4 E4B on MLVU video, 6.9 mWh per inference, and a GPU left completely idle during inference.

VLM On-Device Inference Qualcomm Reka Multimodal Edge AI Vision-language models Edge Inference

Category: perception

Author: @RekaAILabs

Date: 2026-09-22T00:00:00

Duration: 6.506s

Source: https://x.com/RekaAILabs/status/2102436280936395030

Reference: https://reka.ai/labs/research/reka-edgeq

Related Videos

CHC5 Open Camera: FPGA Vision Processing at 2,000 FPS

CHC5 Open Camera: FPGA Vision Processing at 2,000 FPS

@Alacritic_Super · 376 views · 2026-08-31

NanoOWL: Zero-Shot Real-Time Object Detection by NVIDIA

NanoOWL: Zero-Shot Real-Time Object Detection by NVIDIA

@simplifyinAI · 172 views · 2026-08-24

NVIDIANanoOWLZero-Shot

Open-set object detection recognize unseen objects

Open-set object detection recognize unseen objects

@ctorobotics · 146 views · 2026-07-26

PerceptionOpen SetDetection

Ming Image-0.1 Design: Open 6B Image Model for Text-Rich Design

Ming Image-0.1 Design: Open 6B Image Model for Text-Rich Design

@AdinaYakup · 10 views · 2026-09-22

Image GenerationText RenderingRGBA

Tripo P2.0 x GPT-6 Astra: Second-Pass Character Refinement

Tripo P2.0 x GPT-6 Astra: Second-Pass Character Refinement

@Maoku · 116 views · 2026-09-22

3D GenerationTripoGPT-6 Astra

A Drone That Flies on Two Cameras and an IMU, No Map or GNSS

A Drone That Flies on Two Cameras and an IMU, No Map or GNSS

@tokufxug · 76 views · 2026-09-22

DronesVisual navigationStereo Camera