synthetic-data

NVIDIA has released over 10 trillion pre-training tokens and millions of post-training samples as part of its Nemotron open data initiative for building AI agents. The release includes the Nemotron Post-Training v3 Prompt Atlas, an interactive visualization tool, and Nemotron-Personas dataset representing 2.4 billion people across 10 countries.

NVIDIA demonstrated that task-seeded synthetic Q&A data improves model performance across multiple benchmarks in a 100B-token continuation experiment on Nemotron-3 Nano. The approach improved GPQA scores by +11.1 points, MMLU-Pro by +1.8, average code by +1.9, and commonsense understanding by +1.6.

Arcee AI has released Trinity-Large-Thinking, a 400-billion-parameter open-weight reasoning model with a mixture-of-experts architecture that activates only 13 billion parameters per token. The model matches Claude Opus 4.6 on agent benchmarks like Tau2 and PinchBench but lags on general reasoning tasks. The company spent approximately $20 million—roughly half its total venture capital—to train the model on 2,048 Nvidia B300 GPUs over 33 days.

AI2 is developing physical AI systems trained primarily on virtual simulation data rather than expensive real-world demonstrations. The approach, demonstrated through projects like MolmoBot, addresses the historical bottleneck of manually collecting hardware training data.