Flora

Overview

Flora is NVIDIA‘s open-weight data-labeling system for robotics video (temporal segmentation plus captioning), released on 2026-10-25 with weights on HuggingFace under a non-commercial research license. Nvidia’s own evaluation reports approximately $0.72/hr versus approximately $88.50/hr for GPT-6 Astra via API — a stated up to 125× cost ratio on their internal video-annotation benchmark.

Timeline

• 2026-10-27-AI-Digest — Flora open-sourced on 2026-10-25 as an open-weight temporal-segmentation + captioning stack explicitly aimed at robotics-training video annotation. Load-bearing softener: the 125× headline is a Nvidia-reported projection against an assumed $10 GPU-hour rental, scoped to the specific annotation workload benchmarked, and the non-commercial license means this isn’t price-to-ship for a commercial product — the number people will quote is 125×, the number worth remembering is annotation workload only, Nvidia-reported.

Key Developments

1. Open-weight robotics-video labeling system, released October 25, 2026: Flora targets the specific temporal-segmentation + captioning workload that frontier VLMs (notably GPT-6 Astra) have been handling at API rates. The open-weight release puts a non-commercial-licensed alternative on HuggingFace with a dedicated benchmark pitch — the robotics-training-cost arc now has an open-weight hook.

2. Nvidia-reported 125× cost ratio against GPT-6 Astra via API, assumption-dependent: $0.72/hr vs $88.50/hr on Nvidia’s own benchmark, assuming a $10 GPU-hour rental and scoped to the specific annotation workloads measured — not a general video-generation claim and not reproduced by a third party. The comparison axis (temporal segmentation plus captioning versus a frontier VLM API) is a reasonable workload to measure, but practitioners should carry the Nvidia-reported qualifier.

Related

See also: NVIDIA, Astra, Nemotron 3.5 Super VL, Nemotron Bolt, MOC - Open Source Models, MOC - AI Infrastructure.