# NVIDIA's Nemotron TwoTower mixes diffusion and Mamba

> A new 30B mixture-of-experts base model activates just 3B parameters per token and pairs a hybrid diffusion/Mamba design.

Published by The Open Weights on Apr 11, 2026. Canonical: https://theopenweights.com/news/nemotron-twotower-30b-a3b-d2oa

## Key facts

- Company: NVIDIA
- Model: Nemotron TwoTower 30B-A3B Base
- Version: 1
- Category: Text / LLM
- Modalities: Text / LLM
- License: Other (Open weights, commercial use allowed)
- Parameters: 30B (mixture of experts)
- Active parameters: 3B
- Precision: BF16
- Languages: en, es, fr
- Significance: notable
- Published: 2026-04-11
- Last verified: 2026-06-30
- Hugging Face: https://huggingface.co/nvidia/Nemotron-TwoTower-30B-A3B-Base-BF16
- Canonical URL: https://theopenweights.com/news/nemotron-twotower-30b-a3b-d2oa

NVIDIA has published Nemotron TwoTower 30B-A3B Base, an open-weights language model that leans on a few unusual architectural choices at once. According to its [Hugging Face repository](https://huggingface.co/nvidia/Nemotron-TwoTower-30B-A3B-Base-BF16), the model is a mixture-of-experts system with 30 billion total parameters but only about 3 billion active per token — the "A3B" in its name — released in BF16 precision.

The more distinctive part is the "TwoTower" design, which the release describes as a hybrid combining diffusion and Mamba-style components. Mamba is a state-space approach that offers an alternative to standard attention, while diffusion-based text generation departs from the autoregressive decoding most large language models rely on. Bringing both into a single MoE base model is an experimental direction rather than a conventional transformer stack.

## Why it matters

MoE models like this one are attractive because they decouple capacity from compute cost: you get the knowledge capacity of a larger network while only paying to run a small slice of it per token. Pairing that with state-space and diffusion ideas signals NVIDIA's continued willingness to test architectures beyond the standard playbook.

A few practical notes from the record:

- This is a base model, not an instruction-tuned chat variant
- Listed language coverage is English, Spanish, and French
- It ships under a custom ("other") license, so teams should read the terms before commercial use

Context length isn't specified in the release metadata, and as a base checkpoint it's aimed at researchers and builders who want to fine-tune or study the architecture rather than drop it into production as-is. For anyone tracking how hybrid designs perform at scale, it's a concrete artifact to probe.

## Get the model

- [Hugging Face](https://huggingface.co/nvidia/Nemotron-TwoTower-30B-A3B-Base-BF16)

## Sources

- [nvidia/Nemotron-TwoTower-30B-A3B-Base-BF16](https://huggingface.co/nvidia/Nemotron-TwoTower-30B-A3B-Base-BF16) — Hugging Face, Apr 11, 2026

---
Source: The Open Weights (https://theopenweights.com/). Aggregated and written by Claude, curated by humans. Cite as: "NVIDIA's Nemotron TwoTower mixes diffusion and Mamba", The Open Weights, Apr 11, 2026, https://theopenweights.com/news/nemotron-twotower-30b-a3b-d2oa