Nemotron 3 Ultra
StandardNVIDIA's open frontier reasoning model for orchestration, coding, and deep research.
550B-parameter sparse MoE with 55B active parameters and a hybrid Transformer-Mamba design. Built for enterprise agents, coding automation, long-form analysis, and complex tool workflows with a large working context.
▼
Nemotron 3 Ultra is somewhat expensive at $0.60/M tokens input and $3.60/M tokens output. Cached input reads are priced at $0.20/M tokens, making repeated context cheaper. The model supports extended reasoning, a 512K context window.
Benchmarks
This model hasn't been evaluated by Artificial Analysis yet. Benchmark scores will appear once they're published.
Specifications
Performance
Providers
Misc
Compare with
GPT 5.6 Solvs Nemotron 3 UltraGPT 5.6 Lunavs Nemotron 3 Ultra
Claude Fable 5vs Nemotron 3 UltraClaude Opus 4.8vs Nemotron 3 UltraClaude Sonnet 5vs Nemotron 3 UltraClaude Haiku 4.5vs Nemotron 3 Ultra
Gemini 3.6 Flashvs Nemotron 3 UltraGemini 3.5 Flashvs Nemotron 3 UltraGemini 3.5 Flash-Litevs Nemotron 3 UltraGemma 4 31Bvs Nemotron 3 Ultra
Grok 4.5vs Nemotron 3 UltraGrok Build 0.1vs Nemotron 3 UltraGrok 4.1 Fastvs Nemotron 3 Ultra
Muse Spark 1.1vs Nemotron 3 Ultra
DeepSeek V4 Provs Nemotron 3 UltraDeepSeek V4 Flashvs Nemotron 3 Ultra
MiniMax M3vs Nemotron 3 Ultra
GLM 5.2vs Nemotron 3 Ultra
Qwen3.7 Maxvs Nemotron 3 UltraQwen3.7 Plusvs Nemotron 3 UltraQwen3 Coder Nextvs Nemotron 3 Ultra
Mistral Small 4vs Nemotron 3 UltraFrequently Asked Questions about Nemotron 3 Ultra▼
When was Nemotron 3 Ultra released?
▼
Nemotron 3 Ultra was released on June 4, 2026.
Who created Nemotron 3 Ultra?
▼
Nemotron 3 Ultra was created by NVIDIA.
How much does Nemotron 3 Ultra cost?
▼
Nemotron 3 Ultra costs $0.60/M input tokens and $3.60/M output tokens.
What is Nemotron 3 Ultra API pricing?
▼
The Nemotron 3 Ultra API is priced at $0.60 per million input tokens and $3.60 per million output tokens. You can access Nemotron 3 Ultra through idapt.app alongside 200+ other AI models in one workspace.
Is Nemotron 3 Ultra a reasoning model?
▼
Yes, Nemotron 3 Ultra is a reasoning model. It can think step-by-step through complex problems before providing an answer, often yielding better results on difficult tasks like math, coding, and multi-step logic.
Does Nemotron 3 Ultra support image or vision input?
▼
No, Nemotron 3 Ultra does not currently support image or vision input. It is a text-only model.
Does Nemotron 3 Ultra support audio?
▼
No, Nemotron 3 Ultra does not support audio input.
What is the context window of Nemotron 3 Ultra?
▼
Nemotron 3 Ultra has a context window of 512K tokens, meaning it can process approximately 384,216 words of context at once.
Is Nemotron 3 Ultra open source?
▼
Nemotron 3 Ultra is a proprietary model developed by NVIDIA. The model weights and training data are not publicly available.
Where can I use Nemotron 3 Ultra?
▼
You can chat with Nemotron 3 Ultra on idapt.app — no separate API key required. idapt is an AI workspace with 200+ models, agents, computers, and tasks. You can also access Nemotron 3 Ultra through NVIDIA's own API or platform.