NVIDIA
NVIDIA
Language model · United States ·Jun 2026

NVIDIA Nemotron 3 Ultra

Language Open Weights
550B
Parameters
1M
Context
8
Benchmarks

About NVIDIA Nemotron 3 Ultra

NVIDIA Nemotron 3 Ultra is an AI model developed by NVIDIA, in the language category, released in 2026, made available as an open-weights model with 550B params. Nemotron 3 Ultra is NVIDIA's most capable open model, a 550B-total / 55B-active hybrid Mamba-Attention Mixture-of-Expert.

On this page you'll find NVIDIA Nemotron 3 Ultra's full specifications, including a context window of 1M tokens, 8 benchmark results. Review provider pricing and benchmark scores below, or compare NVIDIA Nemotron 3 Ultra head-to-head with other language models.

→ More models by NVIDIA → All language models → AI benchmarks leaderboard

Links

Paper ↗

Run NVIDIA Nemotron 3 Ultra locally — what it takes

NVIDIA Nemotron 3 Ultra (550B) is too large to run on a single consumer or Apple machine. At Q4 its weights alone are about 308 GB — it needs a multi-GPU / datacenter setup:

  • ≈ 2× AMD Instinct MI325X
  • ≈ 2× NVIDIA B200

Lower quantization (Q3/Q2) or CPU + system-RAM offload (e.g. ktransformers) reduces the requirement at the cost of speed.

Software support

·ExLlamaV2 ·KoboldCpp ·LM Studio ·MLX ·Ollama ·SGLang ·TensorRT-LLM ·llama.cpp ·vLLM ?KTransformers ?text-generation-webui

✓ measured · · compatible · — not supported. Informational only — speed is hardware-based.

Community

Would you run NVIDIA Nemotron 3 Ultra again?
No votes yet — be the first.
Did it run? — community reports by hardware

No reports yet. Own this model on some hardware? Be the first to confirm it runs.

Discussion

No comments yet. Start the discussion.

NVIDIA Nemotron 3 Ultra benchmark scores▶
Compare on the benchmark table →
Benchmark Category Score Variant
BrowseComp agentic 44.4%
Terminal-Bench 2.1 agentic 56.4%
LiveCodeBench code 89%
SWE-bench Verified code 70.7%
SciCode code 44.6%
MMLU-Pro general 86.8%
MultiChallenge general 63.8%
Humanity's Last Exam reasoning 26.7%

More models like NVIDIA Nemotron 3 Ultra

AI
DMind-1
 AI
COLLECTIVE-1
 AI
Cogito 1 series
 AI
Euler (Euler大模型)
· 2.5B
 AI
SmolLM-1.7B
· 1.7B
 AI
Llama-3-70B-Synthia-v3.5
· 70B

Frequently asked questions

What is NVIDIA Nemotron 3 Ultra? ▶

Nemotron 3 Ultra is NVIDIA's most capable open model, a 550B-total / 55B-active hybrid Mamba-Attention Mixture-of-Expert It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created NVIDIA Nemotron 3 Ultra? ▶

NVIDIA Nemotron 3 Ultra was developed by NVIDIA and released in 2026.

Is NVIDIA Nemotron 3 Ultra open source or proprietary? ▶

NVIDIA Nemotron 3 Ultra ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.

What is NVIDIA Nemotron 3 Ultra's context window? ▶

NVIDIA Nemotron 3 Ultra supports a context window of 1M tokens, and has 550B params.

How much does NVIDIA Nemotron 3 Ultra cost? ▶

Pricing for NVIDIA Nemotron 3 Ultra depends on the provider. See the providers table on this page for the latest API rates.

How does NVIDIA Nemotron 3 Ultra perform on benchmarks? ▶

NVIDIA Nemotron 3 Ultra is benchmarked across 8 evaluations on GenAIList, including BrowseComp (44.4%). See the full benchmark table below and compare it with other models.

Nemotron 3 ULTRA First Look & Test
Nemotron 3 Ultra and the Open Model Landscape | Nemotron Labs
Nemotron 3 Ultra - NVIDIA's Most Powerful Open Model
NVIDIA's Nemotron 3 Is... Awesome?
NVIDIA Nemotron 3 Ultra 550B MoE Coding Test

Know a great video about NVIDIA Nemotron 3 Ultra?