NVIDIA: Nemotron 3 Nano 30B A3B
🧠 AI Modelnvidia
Efficient 30B MoE model with 3B activated, 262k context for agentic AI.
NVIDIA Nemotron 3 Nano 30B A3B is a Mixture-of-Experts (MoE) language model with 30B total parameters and 3B activated parameters per token, making it one of the most compute-efficient models for its size. It features a context length of 262,144 tokens, enabling long-form reasoning and complex agentic tasks. The model supports text input and output, and includes features such as frequency penalty, logit bias, reasoning tokens, and repetition penalty. It is fully hosted on OpenRouter by NVIDIA, with pricing at $0.05 per million input tokens and $0.20 per million output tokens. Benchmarks show strong performance on reasoning and agentic benchmarks, making it suitable for developers building specialized AI agents.
💡Highlights
- ├─30B MoE, 3B activated per token
- ├─262,144 token context length
- └─Optimized for agentic AI
🎯For
- ├─AI developers
- ├─Agent engineers
- └─NVIDIA ecosystem users