Speed
Output tokens / second
Nemotron 3 Nano 30B A3B overview
Nemotron 3 Nano 30B A3B by NVIDIA
A sparse mixture-of-experts language model that activates just 3B parameters per token while drawing on 30B of learned knowledge, built from the ground up for agentic AI systems, production RAG pipelines, and long-context reasoning at scale.
Nemotron 3 Nano 30B A3B specs
- Model ID
nvidia/nemotron-3-nano-30b-a3b- Developer
- NVIDIA
- Context window
- 262,144 tokens
- Max output
- 228,000 tokens
- Input
- Text, Image
- Output
- Text
- API
- OpenAI-compatible, base URL
https://api.aimlapi.com/v1
What Is Nemotron 3 Nano 30B A3B?
30B total parameters, A3B meaning roughly 3 billion are activated per inference pass. NVIDIA built this model from scratch, not fine-tuned on top of someone else's base, training on 25 trillion tokens of text covering code, math, science, general knowledge, and multilingual data.
What makes it genuinely different from most open-weight models in this size class is the architecture. Rather than stacking standard transformer attention layers throughout, NVIDIA combined three distinct layer types, Mamba-2, Mixture-of-Experts, and grouped-query attention, into a hybrid stack that runs significantly faster in practice while handling very long contexts without the usual memory blowup.
Architecture: Hybrid Mamba-Transformer MoE
The architectural backbone of Nemotron 3 Nano is what separates it from straightforward scaled-down transformers. NVIDIA calls it a Hybrid Mamba-Transformer MoE, and understanding each component helps clarify exactly why the model performs the way it does.
| Parameter | Value |
|---|---|
| Total Layers | 52 |
| Mamba-2 Layers | 23 |
| MoE Layers | 23 |
| GQA Attention Layers | 6 (2 groups) |
| Experts per MoE Layer | 128 routed + 1 shared |
| Activated Experts per Token | 6 |
| Hidden Dimension | 2,688 |
| Activation Function | Squared ReLU (ReLU²) |
| Normalization | RMSNorm |
| Context Window | Up to 262,144 tokens (1M supported) |
Benchmark Performance
Numbers matter more than marketing, so here's where Nemotron 3 Nano 30B A3B actually lands against its direct open-weight competitors, Qwen3-30B-A3B and GPT-OSS-20B.
| Metric | Value |
|---|---|
| AIME 2025, Math Reasoning (no tools) | 89.1% |
| Qwen3-30B-A3B (baseline) | 85.0% |
| GPT-OSS-20B | 91.7% |
| AIME 2025, Math Reasoning (with Python tools) | 99.2% |
| Performance Insight | Near theoretical ceiling with tool integration |
| Inference Throughput vs Qwen3-30B-A3B (H200, 8K→16K) | 3.3× faster |
| Throughput vs GPT-OSS-20B | 2.2× faster (same hardware) |
On long-context evaluations (RULER benchmark), Nemotron 3 Nano outperforms both Qwen3-30B-A3B-Instruct-2507 and GPT-OSS-20B across varying context lengths, a direct payoff from the Mamba-2 linear-time design. FP8 quantization retains approximately 99% of BF16 accuracy, meaning the model runs efficiently on 24GB VRAM cards like the RTX 4090 without a significant quality hit.
Capabilities and Use Cases
Nemotron 3 Nano was fine-tuned and RL-trained with production use cases in mind, not benchmark chasing. It handles both reasoning and non-reasoning modes via a flag in the chat template, useful for cutting latency on simpler tasks while retaining full chain-of-thought quality where it matters.
Agentic Systems
Explicitly designed for multi-step agent loops: tool calling, planning, structured output generation, and code execution. The AIME-with-tools score of 99.2% illustrates how well it integrates with external functions.
RAG & Long-Context Processing
A 262K-token context window means large codebases, legal documents, research papers, or extended session histories fit in a single call. Long-range coherence is maintained without the memory overhead of pure-attention models.
Use cases
NVIDIA documents this model as suitable for coding, reasoning, math, and long-context tasks.
It is also intended for AI agents that need high accuracy and efficiency for targeted tasks.
NVIDIA further describes it as a general-purpose reasoning and chat model for instruction-following, chatbots, RAG systems, and other AI-powered applications.
The model is also documented as useful for developers building AI agent systems and tool-using workflows.
Capabilities
- Function calling
- Streaming
- Structured outputs
- Reasoning
Nemotron 3 Nano 30B A3B benchmarks by 
Artificial Analysis
See how Nemotron 3 Nano 30B A3B compares in generation speed and latency. Data retrieved Nov 3, 2026.
Benchmark results
Compare 6 models available on AI/ML API across 5 independent benchmarks.
Humanity's Last Exam
Expert-written questions across academic subjects.
GPQA Diamond
Graduate-level science questions written to resist web search.
Terminal-Bench 4.0
Agentic tasks completed in a command-line environment.
AA-LCR
Reasoning across long documents of about 100k tokens.
SciCode
Python coding tasks drawn from scientific research.
Nemotron 3 Nano 30B A3B API pricing
| Usage | Price | Unit |
|---|---|---|
| Input | $0.0529 | 1M tokens |
| Output | $0.2116 | 1M tokens |
Prices from the AI/ML API price list, checked on Nov 4, 2026.
Nemotron 3 Nano 30B A3B vs other models: token prices
AI/ML API input and output rates for 5 models, per million tokens.
| Model | Input | Output |
|---|---|---|
 Nemotron 3 Nano 30B A3B | $0.0529 | $0.2116 |
| Nemotron 3.5 Content Safety | $0.2116 | $0.2116 |
| Nemotron 3 Super 120B-A12B | $0.0952 | $0.4761 |
| Nemotron 3.5 Super VL | $0.2645 | $1.59 |
| Nemotron 3 Ultra | $0.6348 | $3.81 |
How to use the Nemotron 3 Nano 30B A3B API
View docs
Build with your agent
Give your agent the model details, code and documentation.
Paste it into your coding agent to build the integration.
1Get your API key
Create API keyexport AIMLAPI_KEY="8f2b4c7d9a1e6b3f…"2Make your first request
curl 'https://api.aimlapi.com/v1/chat/completions' \
-H "Authorization: Bearer $AIMLAPI_KEY" \
-H 'Content-Type: application/json' \
-d '{
"model": "nvidia/nemotron-3-nano-30b-a3b",
"messages": [
{
"role": "user",
"content": "Write a haiku about the ocean."
}
]
}'import os
import requests
headers = {
"Authorization": f"Bearer {os.environ['AIMLAPI_KEY']}",
"Content-Type": "application/json",
}
payload = {
"model": "nvidia/nemotron-3-nano-30b-a3b",
"messages": [
{
"role": "user",
"content": "Write a haiku about the ocean."
}
]
}
response = requests.post("https://api.aimlapi.com/v1/chat/completions", headers=headers, json=payload, timeout=120)
response.raise_for_status()
print(response.json())const response = await fetch("https://api.aimlapi.com/v1/chat/completions", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
"model": "nvidia/nemotron-3-nano-30b-a3b",
"messages": [
{
"role": "user",
"content": "Write a haiku about the ocean."
}
]
}),
});
if (!response.ok) throw new Error(`HTTP ${response.status}`);
console.log(await response.json());Models similar to Nemotron 3 Nano 30B A3B
Related articles
Nemotron 3 Nano 30B A3B FAQ
What is the context window of Nemotron 3 Nano 30B A3B?
Nemotron 3 Nano 30B A3B has a 262,144 tokens context window and can return up to 228,000 tokens.
What inputs and outputs does Nemotron 3 Nano 30B A3B support?
Nemotron 3 Nano 30B A3B takes text as input and returns text.
What is the model ID for Nemotron 3 Nano 30B A3B?
Use nvidia/nemotron-3-nano-30b-a3b as the model id. Requests go to https://api.aimlapi.com/v1/chat/completions.
When was Nemotron 3 Nano 30B A3B released?
Nemotron 3 Nano 30B A3B became available on April 30, 2026.
How much does Nemotron 3 Nano 30B A3B cost?
Nemotron 3 Nano 30B A3B costs $0.0529 input and $0.2116 output per 1M tokens on AI/ML API, billed as you go with no subscription.
Does Nemotron 3 Nano 30B A3B support streaming?
Yes, Nemotron 3 Nano 30B A3B can stream responses as they are generated.
Who created Nemotron 3 Nano 30B A3B?
Nemotron 3 Nano 30B A3B was built by NVIDIA.