NvidiaReleased on Dec 15, 2025

Nemotron 3 Nano (30B A3B): Benchmarks, Pricing & Size

Nemotron 3 Nano (30B A3B) is a language model from Nvidia, released in December 2025.

Nemotron 3 Nano is a 31.6B hybrid MoE model optimized for fast, long‑context agentic reasoning. It mixes Mamba‑2 and Transformer layers with a sparse MoE router (~3.6B active params per token) to deliver up to 4× higher throughput than

Input
Text
Output
Text
Speed
Latency p9558.70s
Throughput P9555.6char/s
Cost
$0.08/ 1M · 8:1 in:out
$0.06 in · $0.24 out

Nemotron 3 Nano (30B A3B) pricing

Providers

Nemotron 3 Nano (30B A3B) starts at $0.0600 per million input tokens and $0.240 per million output tokens via DeepInfra.

ProviderInput $/MOutput $/MMax InputMax OutputLatency p95 sThroughput P95QuantInputOutput
DeepInfra logoDeepInfra
$0.0600$0.240262.1K262.1K
58.70
56 c/s
bfloat16

Nemotron 3 Nano (30B A3B) model size

Nemotron 3 Nano (30B A3B) has 32 billion parameters and was trained on 25 trillion tokens. See how it compares to other models in the same parameter range.

ParametersTraining tokens
32BMoE
25Ttokens
781× tokens-to-params ratio
Large (30–80B)
32B
1B7B70B405B

Nemotron 3 Nano (30B A3B) API

API access coming soon

Nemotron 3 Nano (30B A3B) will be available through our gateway shortly.

Nemotron 3 Nano (30B A3B) examples

Recent arena outputs from Nemotron 3 Nano (30B A3B), picked from the highest-ranked matchups.

Nemotron 3 Nano (30B A3B) license

Nemotron 3 Nano (30B A3B) is released under the NVIDIA Open Model License Agreement license, which permits commercial use, has 32.0B parameters, has a knowledge cutoff of November 2025.

License
NVIDIA Open Model License Agreement
Commercial use allowed
Parameters
32.0B
Knowledge cutoff
November 2025

NVIDIA Open Model License Agreement

FAQ

Common questions about Nemotron 3 Nano (30B A3B).

When was Nemotron 3 Nano (30B A3B) released?

Nemotron 3 Nano (30B A3B) was released on December 15, 2025 by Nvidia. This is the official Nemotron 3 Nano (30B A3B) release date tracked on LLM Stats.

How big is Nemotron 3 Nano (30B A3B)?

Nemotron 3 Nano (30B A3B) has 32 billion parameters. It was trained on 25.0 trillion tokens. It ships as an open-weight model, so you can download and run it on your own hardware.

Who created Nemotron 3 Nano (30B A3B)?

Nemotron 3 Nano (30B A3B) was created by Nvidia.

What is the license for Nemotron 3 Nano (30B A3B)?

Nemotron 3 Nano (30B A3B) is released under the NVIDIA Open Model License Agreement license. This is an open-source / open-weight license that permits self-hosting.

What is the knowledge cutoff date for Nemotron 3 Nano (30B A3B)?

Nemotron 3 Nano (30B A3B) has a knowledge cutoff of November 2025, meaning it was trained on data up to that point and may not know about events after it.

Where can I use Nemotron 3 Nano (30B A3B)?

Nemotron 3 Nano (30B A3B) is available through 1 provider including DeepInfra.

Where is the Nemotron 3 Nano (30B A3B) paper or technical report?

Nemotron 3 Nano (30B A3B) has a paper or technical report available at https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3-Nano-Technical-Report.pdf. Use that source for architecture, training, release and evaluation details.