Ran Zilberstein

Multimodal models can now handle audio natively with improved efficiency, achieving state-of-the-art results in complex tasks like document understanding and agentic computer use.

Nemotron 3 Super proves you can achieve comparable accuracy to existing 120B models, but with significantly higher inference throughput, by combining Mamba, Attention, and Mixture-of-Experts.

You can slash LLM inference costs without sacrificing quality by strategically pruning experts, quantizing, and swapping full attention for windowed attention, as demonstrated on gpt-oss-120B.

Synthetic data can significantly overestimate the real-world throughput gains from speculative decoding, highlighting the critical need for benchmarks like SPEED-Bench that use diverse, production-realistic workloads.