nemotron-3-nano:30b

Upstream reference: Nemotron 3 Nano BF16

Publisher: NVIDIA

NVIDIA describes a hybrid Mamba-2, attention and mixture-of-experts text model with configurable reasoning through the chat template.

Publisher-described context, separate from Sigma5C observations and recorded catalog statistics.

Published upstream specifications

Total parameters

30 billion Source

Active parameters reported in card

3.5 billion Source

Named checkpoint precision

BF16 Source

Upstream license metadata

NVIDIA Nemotron Open Model License Source

Publisher-described uses

• Text assistants, reasoning and specialized agent applications. Source

Reference limitations

• The A3B name is not the card’s precise active-parameter count. BF16 upstream weights do not establish the precision of a local or cloud conversion. Source

A recorded tag and an upstream card association do not verify the deployed checkpoint, quantization, configured context or current hosted availability.

Publisher sources

• Nemotron 3 Nano BF16 official documentation — NVIDIA. Checked 2026-11-02.

Links

Recorded references do not establish downloadable weights, licensing or current availability.

• NVIDIA — official site
• Ollama
• Provider page
• Hugging Face
• Official model announcement / documentation

Where nemotron-3-nano:30b comes from

Named in 2 observation pages

• 14B-32B local models — all 47 · V1
• Pre-retirement tool-use observation: three models through Ollama cloud · V1