Nemotron 3 Ultra
NVIDIA Nemotron 3 Ultra is a large-scale reasoning and orchestration model built on a hybrid Transformer-Mamba mixture-of-experts architecture. It is designed to handle complex text-based tasks with a focus on advanced reasoning and tool interaction.
Nemotron 3 Ultra specs and pricing
- Provider: NVIDIA
- Released: June 4, 2026
- Context window: 262K tokens
- Max output: 183K tokens
- Input: text
- Output: text
- Price: $0.60 input / $2.40 output per 1M tokens
- Reasoning: yes
- Tool calling: yes
What Nemotron 3 Ultra is good at
- Complex logical reasoning tasks
- Orchestrating multi-step workflows
- Processing large volumes of text data
To-dos to try with Nemotron 3 Ultra
- Analyze a complex technical document and outline the logical dependencies between each section.
- Draft a multi-step project plan and identify the specific tools needed to execute each phase.
- Summarize a long-form text and extract key arguments for a structured report.
FAQ
- What is the primary architecture of this model? It utilizes a hybrid Transformer-Mamba mixture-of-experts architecture with 55B active parameters out of 550B total.
- Can this model interact with external tools? Yes, the model supports tool calling capabilities to assist in orchestration tasks.
- How much does Nemotron 3 Ultra cost? Nemotron 3 Ultra costs $0.60 per million input tokens and $2.40 per million output tokens through OpenRouter. In todo.is a typical agent step costs about 2.02 credits ($1 = 100 credits).
- What is the context window of Nemotron 3 Ultra? Nemotron 3 Ultra can read up to 262K tokens at once and write up to 183K tokens in one reply.
- Can I use Nemotron 3 Ultra without coding? Yes. In todo.is you pick Nemotron 3 Ultra 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.