Nemotron Nano 9B v2 is a 9 billion parameter model with a hybrid architecture built primarily from Mamba-2 and MLP layers paired with just four Transformer attention layers, designed as a unified model for both reasoning and non-reasoning tasks. Pre-trained on roughly 20 trillion tokens with a September 2024 data cutoff, it was post-trained with synthetic reasoning traces drawn from DeepSeek-R1 and Qwen models alongside reinforcement learning. The model offers controllable extended reasoning through system prompts, letting developers enable reasoning with `/think` or disable it with `/no_think`, with configurable thinking budgets during inference. This makes it well suited to chatbots, retrieval-augmented generation, AI agents, and instruction-following applications that need reasoning without excessive token overhead. It supports 16 languages, including English, German, French, Spanish, Italian, Japanese, Korean, Portuguese, Russian, and Chinese. With a 128K token context window and a compact 9B footprint, it runs efficiently while delivering strong reasoning results, scoring 72.1% on AIME25 and 97.8% on MATH500. It was released in August 2025 under the NVIDIA Open Model License for commercial use, balancing inference speed and capability for production deployments.
Nemotron Nano 9B v2 runs on 1 route through the Opper gateway. Compare residency, ZDR, and training posture at a glance β full data-handling detail per route below.
| Provider | Region | Zero data retention | Training | Input | Output |
|---|---|---|---|---|---|
| US | Enterprise | No | $0.04 | $0.16 |
Training posture across routes: No training on prompts by default.
Each route hosting Nemotron Nano 9B v2 has its own privacy posture, residency, and GDPR terms. Postures are maintained by Opper with a last-verification timestamp.
Zero data retention is available via Opper Enterprise contract. No training on customer data. US; unknown.
Independent benchmark scores β composite indices for reasoning, coding, and math, plus individual eval scores where available.
Call Nemotron Nano 9B v2 through the Opper gateway with one API key. Let your coding agent set it up, or call it directly β Opper is drop-in compatible with the OpenAI, Anthropic, and Google AI SDKs.
Copy this and paste it into your coding agent β Claude Code, Cursor, Codex, and more β and it'll wire up Opper for you.
import OpenAI from "openai";const client = new OpenAI({apiKey: process.env.OPPER_API_KEY,baseURL: "https://api.opper.click/v3/compat",});const completion = await client.chat.completions.create({model: "deepinfra/nvidia/NVIDIA-Nemotron-Nano-9B-v2",messages: [{ role: "user", content: "Hello" }],});console.log(completion.choices[0].message.content);
Side-by-side on privacy, EU hosting, pricing, and benchmarks.
