NVIDIA Releases Nemotron 3.5 Super VL, an Open 120B Multimodal MoE Model
Summary
NVIDIA releases Nemotron 3.5 Super VL, an open 120-billion-parameter mixture-of-experts multimodal model that activates 12 billion parameters per token for agentic work. Scale AI says post-training it for Skydance credit processing cuts inference cost 40 times. Ramp claims 10x lower latency for receipt parsing.
Key Points
- NVIDIA says Nemotron 3.5 Super VL scores 63.44% on Hermes Bench versus 56.33% for Nemotron 3 Ultra.
- AI21 says post-training the model for threat intelligence lifts accuracy 41%.
- NVIDIA reports LiLiCorr speculative decoding on DGX Spark delivers up to 2.3x faster token generation.