Nemotron Nano 9B V2
NVIDIA Nemotron Nano 9B V2 is NVIDIA's unified reasoning and non-reasoning model with /think and /no_think controls, tool calling, and 128K context.
Model Overview
Capabilities, design details, and architectural traits
Nemotron Nano 9B V2 - Unified reasoning and non-reasoning model
Nemotron Nano 9B V2 is NVIDIA's model for both reasoning and non-reasoning tasks. It is built to generate a reasoning trace first, then a final response, with runtime control over that behavior.
| Distinctive trait | What is documented |
|---|---|
| Unified task mode | Designed as a unified model for both reasoning and non-reasoning tasks. |
| Reasoning trace control | Uses /think and /no_think style control for reasoning on or off. |
| Tool calling support | Documented with native tool-calling use. |
| Long context | Supports up to 128K context. |
Reasoning pattern
The model can answer with intermediate reasoning traces or without them, depending on the prompt control. Its documented behavior is to produce the trace first and the final response afterward. The model card also describes a runtime thinking budget for controlling how much it reasons.
Architecture focus
The model uses a hybrid Mamba-2 and MLP design with four attention layers. The official docs place it in the Nemotron Nano v2 family and describe it as optimized for long-sequence use.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | NVIDIA | Anthropic | Anthropic |
| Release Date | August 18, 2025 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | Mar 2025 | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 128K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.05 Best Input Pricing | $4 | $2 |
| Output Pricing | $0.20 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | text | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 6.8 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | - | - | - |
| Agentic Index | - | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from NVIDIA
Other models by NVIDIA
Top AI Models
Leading alternatives by intelligence score
