Pyyan / Open-Weight Models / Nemotron 3.5 Lightning

Nemotron 3.5 Lightning

Hybrid Mamba and transformer, 3B active, and the training data ships with it.

currentNVIDIAVerified 2026-09-02

30B

MoE, 3B active

Specifications

Context128Ksource · as of 2026-09-02 · medium confidence
Input ($/Mtok)Weights onlysource · as of 2026-09-02
Output ($/Mtok)Weights onlysource · as of 2026-09-02
LicenceOpenMDW-1.1source · as of 2026-09-02
Max output32Ksource · as of 2026-09-02 · low confidence
Parameters30B total, 3B activesource · as of 2026-09-02
Released11 Aug 2026source · as of 2026-09-02
Model IDnvidia/nemotron-3.5-lightningsource · as of 2026-09-02 · medium confidence

What matters

  • 30B mixture of experts with roughly 3B active per token, hybrid Mamba and transformer
  • Weights, training data and recipes all released, under NVIDIA's OpenMDW-1.1 licence
  • Scores 24 on the Artificial Analysis Intelligence Index, up from 15 for Nemotron 3 Nano

Best for long-running agents.