NVIDIA Launches Nemotron 3.5 Lightning: A 30B MoE Model Built for Always-On Agents
kuchaev · x · 2026-08-11
NVIDIA has introduced Nemotron 3.5 Lightning, a 30B hybrid Mamba-Transformer MoE model with 3B active parameters, distilled from Nemotron 3 Ultra.
Built specifically for "always-on" agents to handle high-volume, specialized tasks, it delivers up to 4x the output speed of similar-sized models. Key features include:
- Three built-in speculators: MTP, DFlash, and DSpark
- Trained for agent harnesses including coding, tool use, and multi-turn interactions
- Up to 1M-token context length, with BF16 and NVFP4 precision support
- Runs across Jetson, DGX Spark, Hopper, and Blackwell architectures
Inference framework SGLang has already announced Day 0 support for the model.
Related event: NVIDIA Unveils Nemotron 3.5 Lightning and NeMo Switchyard Router(55 posts)→
More from Infra
- Jensen Huang: Musk's AI buildout moved from K's to M's, now a CSP-tier NVIDIA customer — elonmusk · 2026-10-16
- EpochAI pegs GPT-6 training at up to 1.8e27 FLOP; scaling01 calls param estimates way too high — scaling01 · 2026-10-16
- Meta's Muse Spark Open scores 39 vs 48 for closed Spark 1.3 on AA Intelligence Index — EAccelerate_42 · 2026-10-16
- LLM spots useless 95GB-per-frame memcpy in ffmpeg dblur, cutting runtime from 2452ms to 49ms — IgorCarron · 2026-10-16
- DeepSeek-V4-Flash on 4x Radeon AI PRO R9700: the 'refused' 4-card config hits ~79 tok/s — brainchillzZ · 2026-10-16
- Tree packing lands in TRL's AsyncGRPO, cutting agentic RL training compute by over 26x — joecole · 2026-10-16