Models · MarkTechPost ·
NVIDIA Releases Nemotron-Labs-3-Puzzle-75B-A9B: A Compressed Hybrid MoE LLM Delivering 2.03x Server Throughput at Matched User Throughput
NVIDIA released Nemotron-Labs-3-Puzzle-75B-A9B, a compressed version of Nemotron-3-Super that uses hardware-aware structural compression and short distillation recovery phases. The model reduces total and active parameters while improving server throughput and increasing token concurrency on supported GPUs.