---
title: "Nemotron Ultra 253B — LLM-TEST.com"
url: /models/nemotron-ultra-253b/
canonical_url: https://llm-test.com/models/nemotron-ultra-253b/
language: en
type: model
summary: "NVIDIA's largest open-weight dense model, a reasoning-focused flagship served through NVIDIA NIM and tuned to run efficiently on its own accelerators."
last_updated: "2026-11-05"
---

# Nemotron Ultra 253B

_NVIDIA · Reasoning · open weights_

NVIDIA's largest open-weight dense model, a reasoning-focused flagship served through NVIDIA NIM and tuned to run efficiently on its own accelerators.

## At a glance

- Released: Mar 5, 2026
- Provider: NVIDIA
- Context window: 256,000 tokens
- Max output: 32,000 tokens
- Modalities: Text → text
- Knowledge cutoff: December 2025
- API model ID: `nemotron-ultra-253b`
- License: Open weights (NVIDIA Open Model License)

## Pricing

- Standard: $3.50 input / $14 output per 1M tokens (aggregator estimate)
- Example: a 1,500-token prompt + 650-token answer ≈ $0.014 (standard)

## Best use cases

- **Advanced reasoning** — Tackles math, science, and multi-step planning problems with long deliberation.
- **On-prem deployment** — Runs open weights inside regulated or air-gapped NVIDIA environments.
- **Synthetic data** — Generates high-quality training data and reasoning traces for smaller models.

## Strengths

- Strong open-weight reasoning
- Optimized for NVIDIA hardware
- Long context

## Limitations

- Heavy and costly to serve
- Text only

## Where to get access

- API: build.nvidia.com (NVIDIA NIM) — https://build.nvidia.com
- Also available: downloadable from NGC and Hugging Face
- Docs: https://docs.api.nvidia.com

Full comparison and live pricing: https://llm-test.com/ · directory: https://llm-test.com/models/