Nvidia open-sources Flora robot-video data engine
What you need to know
Nvidia has open-sourced Flora, a data engine that labels robot video using three fine-tuned Qwen3.5 models: a 9B segmenter and two 27B models, including Flora-Rewarder. Nvidia estimates labeling costs about 72 cents per hour, versus roughly $88.50 with GPT-6 Astra. Those cost figures are projections based on an assumed $10 per GPU hour. Nvidia says Flora already annotates up to 200,000 hours of video per day across 512 GB300 GPUs during GR00T training. On the public WGO-Bench, Flora leads on action boundaries but trails Astra overall and Gemini 3.8 Flash on description accuracy.
Generated by AI from the reporting · updated 3 hours ago
Timeline
Follow the reports to see every side of the story.
- The DecoderNvidia open-sources Flora to make robot video labeling 125x cheaper than GPT-6 Astra
Nvidia open-sourced Flora, an open data engine that labels robot video for about 72 cents per hour versus roughly $88.50 with GPT-6 Astra by Nvidia's estimate. Flora consists of three fine-tuned Qwen3.5 models, a 9B segmenter and two 27B models including Flora-Rewarder, and Nvidia says it already annotates up to 200,000 hours of video per day across 512 GB300 GPUs during GR00T training. The cost figures are projections based on an assumed $10 per GPU hour, and on the public WGO-Bench Flora leads on action boundaries but trails Astra overall and Gemini 3.8 Flash on description accuracy.
Heat trend for this story
Not enough continuous data yet to chart a trend.