Nvidia Open-Sources Flora for Robot Data Labeling
Nvidia launched its open-source data engine Flora, which uses three fine-tuned Qwen3.5 models to label robot videos at 125x lower cost than GPT-6 Astra. The system excels at detecting action errors, positioning it as a key part of Nvidia's strategy to control the robotics value chain.
Summarized by AI from one outlet
Source
- THE DECODERRead original ↗
Latest
- Open source
OpenMobile-2: Open-Source Phone Agents Rival Proprietary Models
Researchers released OpenMobile-2, an open-source smartphone agent model. Trained on an environment with over 110 apps, it performs on par with proprietary models on real Android devices, suggesting breadth of trained apps matters more than data volume.
- Open source
BFL Open-Sources FLUX 3 Image Klein
Black Forest Labs released the open-weights model FLUX 3 Image Klein, featuring a 5.6B parameter architecture. The model supports native 4K and text rendering, with its distilled version generating images in about 8.9 seconds on an RTX 4070 GPU and integrating with ComfyUI.
- Open source
Microsoft open-sources Agensh multi-agent harness
Microsoft Research open-sourced Agensh, a decentralized multi-agent harness. It scales to 1,024 agents on the ProgramBench benchmark, offering a self-organizing collaboration model without a central orchestrator.
- Open source
Cognition Open-Sources Devin Security Plugin
Cognition released the MIT-licensed Devin Vulnerability Fixes plugin featuring three agent skills. The tool inventories, patches, and tests PHP, Node.js, and self-managed GitLab deployments affected by recent critical RCE flaws, opening one pull request per repository.
- Open source
Open-source tool GPU Unlock unlocks AI performance on RTX 30 series
The open-source project GPU Unlock removes artificial limits on Ampere architecture cards without changing drivers or VBIOS. It nearly doubles tensor core throughput under specific loads, improving AI compute performance.
All news → · All rights belong to the original creators