According to emollick, firms face rising costs to fine tune small models as frontier models like GPT6 Astra outpace them for complex tasks.

According to AIatMeta, AIRA3 ranked 8th of ~4,000 in NVIDIA’s Kaggle Nemotron-30B fine-tuning, signaling expert-level targeted capability gains.

According to NVIDIAAI, Nemotron-Labs-TwoTower keeps 98.7% quality while generating 2.42x faster by parallel token writing from a split 30B model.

According to TheRundownAI, Nvidia unveiled RTX Spark PCs, Vera CPU, Cosmos 3 model, and Nemotron 3 Ultra, signaling on-device agent computing.