Nemotron 3 Super
NVIDIA's Nemotron 3 Super is a 120B-parameter hybrid model designed for high-efficiency performance in complex multi-agent tasks. It utilizes a hybrid Mamba-Transformer architecture that activates only a fraction of its parameters to balance accuracy with computational speed.
Nemotron 3 Super specs and pricing
- Provider: NVIDIA
- Released: March 11, 2026
- Context window: 1M tokens
- Max output: 262K tokens
- Input: text
- Output: text
- Price: $0.30 input / $2.40 output per 1M tokens
- Reasoning: yes
- Tool calling: yes
What Nemotron 3 Super is good at
- Complex multi-agent system coordination
- Tasks requiring high reasoning capabilities
- Large-scale text processing and analysis
To-dos to try with Nemotron 3 Super
- Analyze a complex multi-step project plan to identify potential bottlenecks between different agent roles.
- Draft a detailed technical report based on a massive set of internal text documents.
- Use tool-calling to orchestrate a series of automated actions across a software development workflow.
FAQ
- What is the underlying architecture of Nemotron 3 Super? It is built on a hybrid Mamba-Transformer architecture that uses a mixture-of-experts approach to activate only 12B parameters at a time.
- What types of data can this model process? The model is designed to process and generate text-based input and output.
- How much does Nemotron 3 Super cost? Nemotron 3 Super costs $0.30 per million input tokens and $2.40 per million output tokens through OpenRouter. In todo.is a typical agent step costs about 1.18 credits ($1 = 100 credits).
- What is the context window of Nemotron 3 Super? Nemotron 3 Super can read up to 1M tokens at once and write up to 262K tokens in one reply.
- Can I use Nemotron 3 Super without coding? Yes. In todo.is you pick Nemotron 3 Super as your agent's model and write to-dos in plain words, on the web, the Android app, WhatsApp or Telegram. No API key needed: credits are included, or you can bring your own key.
JavaScript is required to use the todo.is app.