Today for AI
HOT RADAR
opensourceHEAT 11.7°

AI CURATED HOT STORY · 10/25/2026

Nvidia Open-Sources Flora Robot Video Labeling Engine, Cutting Costs by 125x

1 reports archived1 independent sourcesupdated 10/25/2026, 18:08:49
Synthesis & Latest Updates
1 Sources Cross-Validated

Nvidia has open-sourced Flora, a data engine designed for robot video annotation that supports action descriptions, process ratings, and outcome judgments. Compared to using GPT-6 Astra for large-scale labeling, Flora reduces the cost per hour of video from $88 to $0.72, significantly lowering the barrier to acquiring training data for embodied AI.

LATEST/Nvidia has open-sourced Flora, a data engine designed for robot video annotation that supports action descriptions, process ratings, and outcome judgments. Compared to using GPT-6 Astra for large-scale labeling, Flora reduces the cost per hour of video from $88 to $0.72, significantly lowering the barrier to acquiring training data for embodied AI.

Hourly heat curve

2 fully observed hours
Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first
  1. The DecoderT2·78 pts

    Nvidia Open-Sources Flora Robot Video Labeling Engine, Cutting Costs by 125x

    • Flora provides three critical label types simultaneously: temporal action descriptions, step-by-step progress ratings, and task completion judgments, filling a gap left by single-function tools.
    • By optimizing with specialized models, Flora reduces the cost of annotating one hour of robot video to approximately 1/125th of traditional LLM solutions ($0.72 vs $88).
    • The tool aims to accelerate the construction of large-scale robot training datasets, directly competing with rivals like Google's Gemini 4 Argon in the robotics perception space.