Nvidia launched Nemotron 3.5 Lightning, an open-source 30-billion-parameter model based on a mixture-of-experts architecture. Thanks to this design, only 3 billion parameters are active at any given time, which significantly reduces inference costs while maintaining high performance. The company also released NeMo Switchyard, an open-source Rust library that routes each step of an AI workflow to the most suitable model in terms of cost, latency, or capability requirements. Nemotron 3.5 Lightning completes a model family that already includes Nano, Super, and Ultra variants, positioning itself as the option optimized for agentic workloads at massive scale. Both tools were publicly released on August 11, 2026.
Source: Read the original article
