Nemotron 3 Ultra 550B A55B is NVIDIA's largest open-weight reasoning model, with 550B total parameters and 55B active per token through a hybrid Mamba-Transformer LatentMoE architecture with Multi-Token Prediction. It natively supports a 262K token context and speculative decoding, and is built for the most demanding agentic AI workloads. Pre-trained on trillions of tokens spanning code, math, science, and general knowledge across many natural and programming languages, the Ultra model is tuned for long-running agent loops and high-complexity problem-solving. NVIDIA reports substantially higher inference throughput than several similarly scaled competitors on long-context settings, with the LatentMoE design projecting tokens into a smaller latent dimension for more efficient expert routing. The model excels at multi-agent coordination, complex reasoning patterns, and knowledge-intensive retrieval at scale, and is available across NVFP4 and BF16 precisions for different hardware configurations.
Nemotron 3 Ultra 550B A55B runs on 1 route through the Opper gateway. Compare residency, ZDR, and training posture at a glance — full data-handling detail per route below.
| Provider | Region | Zero data retention | Training | Input | Output |
|---|---|---|---|---|---|
| Nebius | US | Enterprise | No | $1.00 | $3.00 |
Training posture across routes: No training on prompts by default.
Each route hosting Nemotron 3 Ultra 550B A55B 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; SCCs; DPA available.
Call Nemotron 3 Ultra 550B A55B through the Opper gateway with one API key. Let your coding agent set it up, or call it directly — Opper is drop-in compatible with the OpenAI, Anthropic, and Google AI SDKs.
Copy this and paste it into your coding agent — Claude Code, Cursor, Codex, and more — 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.ai/v3/compat",});const completion = await client.chat.completions.create({model: "nebius/nemotron-3-ultra-550b-a55b",messages: [{ role: "user", content: "Hello" }],});console.log(completion.choices[0].message.content);
Side-by-side on privacy, EU hosting, pricing, and benchmarks.
