NVIDIA Free 10 tokens / message Balanced

Nemotron 3 Super 120B · NVIDIA

NVIDIA Nemotron 3 Super is a 120B-parameter hybrid Mamba-Transformer MoE model (only 12B active per token) for maximum efficiency and accuracy in complex multi-agent applications. 1M-token context.

Efficient hybrid MoE Multi-agent 1M-token context

Try it free

Hi, how can I help today?
Open in full chat → Compare models side by side, save your sessions and memory

About Nemotron 3 Super 120B

NVIDIA Nemotron 3 Super is a 120B-parameter hybrid Mamba-Transformer MoE model (only 12B active per token) for maximum efficiency and accuracy in complex multi-agent applications. 1M-token context.

Where it shines: Efficient hybrid MoE · Multi-agent · 1M-token context.

How to use Nemotron 3 Super 120B

  1. 1

    Type or upload

    Type what you want in the box above — or upload the file if the tool asks for one.

  2. 2

    Generate

    Click the main button. Wait 2-30 seconds depending on the model and input size.

  3. 3

    Download or share

    Download the result or share the direct link. No watermark, ready to use.

Frequently asked questions

How much does it cost to use Nemotron 3 Super 120B?

Nemotron 3 Super 120B is one of the free models in the catalog. Each use discounts 10 tokens from your pool, but open models like Nemotron 3 Super 120B don’t cost us, so the rate-limit is generous. A free account comes with 500 initial tokens and 25 more every day — you usually don’t get to touch the card.

Is there a limit on the use of Nemotron 3 Super 120B?

There is no fixed monthly fee for Nemotron 3 Super 120B on the free account — the actual limit is the rate per minute/hour, not per month. Anonymous are limited by IP; with account you can do much more volume.If you reach 500+25 tokens and need more, a Pro plan at $9/month covers it.

What makes Nemotron 3 Super 120B special?

NVIDIA fine-tunes its models for fast inference on its own optimized hardware — good at technical questions and reasoning, its specific strengths are: efficient hybrid moe, multi-agent and context 1m tokens.

How fast does Nemotron 3 Super 120B respond?

Nemotron 3 Super 120B has a balanced speed: 5-15 seconds per response — neither the fastest nor the slowest in the catalog.The actual time also depends on the length of the prompt and the load of the datacenter — models with huge context take longer when you enter very long texts.

How do I use Nemotron 3 Super 120B in ia.gratis?

You can use Nemotron 3 Super 120B from /chat/ by selecting Nemotron 3 Super 120B in the picker, or via the REST API with `model=nemotron-3-super` in the body of the POST. Quick summary: 120B hybrid moE with 1M context.The internal model identifier is `nemotron-3-super` — useful when integrating via API.