gekro
GitHub LinkedIn
News

AI News

Nvidia open-sources Flora for robot video labeling at 1/125th GPT-6 Astra cost; Microsoft releases decentralized agent orchestration framework

Nvidia's Flora data engine cuts robot video labeling costs from $88 to $0.72 per hour using fine-tuned open models; Microsoft open-sources Agensh for scaling agent swarms to 1,024 workers.

1 min read 3 sources

Nvidia released Flora, an open-source data engine that labels one hour of robot video for $0.72 instead of $88 when using GPT-6 Astra, while achieving slightly higher quality scores. The system is built on three fine-tuned Qwen3.5 models and detects errors in robot actions more reliably than competing approaches, though it occasionally misses edge cases. (The Decoder)

Microsoft Research open-sourced Agensh, a multi-agent harness that orchestrates up to 1,024 agents without a central coordinator. The framework uses self-organizing workers sharing a Git repository, chat interface, and findings board, scaling to ProgramBench tasks within a 6-hour budget. (MarkTechPost) Separately, Anthropic published research showing single-agent architectures now outperform multi-subagent setups in speed and complexity handling, with a published migration kit available on GitHub to help teams consolidate agent structures. (The Decoder)

Compiled automatically from the linked sources and published without manual editing - a neutral summary of third-party reporting, for information only. Every claim links to its origin. Not original reporting.