NVIDIA Nemotron 3.5 Lightning 30B-A3B

Nvidia llm

Hybrid Mamba-Transformer MoE with MTP, pre-trained in NVFP4 on 20T+ tokens, 1M context, reasoning on/off.

overall ⌄

29.9

116 of 125 ranked

price · $/M tokens

self-hosted

open weights, no first-party API

to run it yourself ⌄

18 GB

minimum · 72 GB recommended

Benchmarks

What its maker published, and what anyone else measured. Every figure links the document it came from.

benchmarkpublishedmeasured
MMLU-Pro81.94 thinking—
GPQA75.44 thinking—
SWE-bench51.56 —
HLE11.72 thinking—
BrowseComp36.97 —

Lineage

No recorded lineage — root or standalone model.

view in full graph →

Metadata

orgNvidia
released2026-08-11
params30B-A3B
licenseopenmdw-1.1
hf idnvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16