NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it...
- Modalities
- text → text
- In / out per 1M
- Free / Free
- Context
- 1M tokens
- Added
- Jun 4, 2026
- Max output
- 66K tokens
- Tokenizer
- Other
- Reasoning
- optional
Pricing
Per 1M tokens. The provider price and our flat 3% fee are separate columns — what you pay is their sum.
| Per 1M tokens | Provider | + 3% fee | You pay |
|---|---|---|---|
| Input | Free | — | Free |
| Output | Free | — | Free |
Providers
One company serves Nemotron 3 Ultra (free). Figures are each provider’s own list price per 1M tokens — your OpenKey price is in the card above.
| Provider | Context | Input /M | Output /M |
|---|---|---|---|
| Nvidia | 1M | Free | Free |
Provider list from the OpenRouter catalog, Jul 28, 2026. All providers
Benchmarks
Artificial Analysis
- Agentic index
- 27.4
- Coding index
- 49.3
- Intelligence index
- 37.8
Design Arena
| Category | Elo | Win rate | Rank |
|---|---|---|---|
| 3D models | 1213 | 43.6% | #40 |
| ASCII art models | 1093 | 35.8% | #44 |
| Code models | 1174 | 37% | #60 |
| Data viz models | 1167 | 38.7% | #63 |
| Game dev models | 1195 | 39.3% | #53 |
| SVG models | 1140 | 38.6% | #45 |
| UI components models | 1180 | 38.7% | #54 |
| Websites models | 1140 | 32.6% | #76 |
Head-to-head preference voting. How we filter and rank
Supported parameters
- include_reasoning
- max_tokens
- reasoning
- seed
- temperature
- tool_choice
- tools
- top_p
Call it
OpenAI-compatible: point your SDK at api.openkey.ai/v1 and use
model nvidia/nemotron-3-ultra-550b-a55b:free.
curl https://api.openkey.ai/v1/chat/completions \
-H "Authorization: Bearer ***" \
-H "Content-Type: application/json" \
-d '{
"model": "nvidia/nemotron-3-ultra-550b-a55b:free",
"messages": [{"role": "user", "content": "Hello"}]
}'import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.openkey.ai/v1",
api_key=os.environ["OPENKEY_API_KEY"],
)
completion = client.chat.completions.create(
model="nvidia/nemotron-3-ultra-550b-a55b:free",
messages=[{"role": "user", "content": "Hello"}],
)
print(completion.choices[0].message.content)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.openkey.ai/v1",
apiKey: process.env.OPENKEY_API_KEY,
});
const completion = await client.chat.completions.create({
model: "nvidia/nemotron-3-ultra-550b-a55b:free",
messages: [{ role: "user", content: "Hello" }],
});
console.log(completion.choices[0].message.content);About Nemotron 3 Ultra (free)
Nemotron 3 Ultra is NVIDIA's largest open Nemotron release: a 550B-parameter mixture-of-experts model that only activates 55B parameters per token, running on a hybrid Transformer-Mamba architecture. It carries a 1M-token context window and optional reasoning (default effort: high). On Artificial Analysis benchmarks it posts a 37.8 intelligence index, 49.3 coding index, and 27.4 agentic index — numbers that put it in the mid-pack for general reasoning but ahead on coding relative to agentic tasks. Output is text-only, with tool calling and adjustable reasoning effort exposed via the API.
This is the free-tier version of nvidia/nemotron-3-ultra-550b-a55b, which normally runs $0.50/$2.20 per million input/output tokens — here both are $0.00. Its 1M-token context window beats 78% of catalog models. At 550B total parameters with 55B active, it's the largest model in NVIDIA's Nemotron 3 lineup, well beyond the 120B Super and 30B Nano variants. Reasoning is optional rather than mandatory, and defaults to high effort.
Questions
- Is Nemotron 3 Ultra (free) free to use?
- Yes — this specific listing (nvidia/nemotron-3-ultra-550b-a55b:free) has $0.00 input and $0.00 output pricing per 1M tokens via OpenKey, no 3% fee applied since there's nothing to mark up. NVIDIA also offers a paid version, nvidia/nemotron-3-ultra-550b-a55b, at $0.50 per 1M input tokens and $2.20 per 1M output tokens if you need higher rate limits or priority access.
- What is Nemotron 3 Ultra's context window?
- It's 1,000,000 tokens, larger than 78% of models in the OpenKey catalog. That's enough to hold very large codebases, lengthy multi-document research sets, or hundreds of pages of text in a single request, with a max completion length of 65,536 tokens per response.
- Does Nemotron 3 Ultra support tool calling and reasoning?
- Yes. It supports tools and tool_choice for function calling, plus an optional reasoning mode with 'high' and 'medium' effort settings — reasoning defaults to high but isn't mandatory, so you can turn it off per request. It also accepts seed, temperature, and top_p for output control.
- How does Nemotron 3 Ultra compare to Nemotron 3 Super?
- Nemotron 3 Ultra is the larger model — 550B total parameters (55B active) versus Nemotron 3 Super's 120B total (12B active). Super is priced at $0.085/$0.40 per 1M input/output tokens (or free in its own :free tier), while Ultra's paid version costs $0.50/$2.20 per 1M tokens — Ultra targets larger context and heavier reasoning workloads, Super is the cheaper mid-tier option.
- Is Nemotron 3 Ultra open source?
- It's published as an open model on Hugging Face under the ID nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16, making weights available for self-hosting or inspection. NVIDIA has built the Nemotron family specifically as open releases alongside its GPU and CUDA platform business.
More from NVIDIA
All NVIDIA models →
NVIDIANemotron 3 Ultra
NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it...
ctx 1M in $0.515 out $2.27
NVIDIAFreeNemotron 3.5 Content Safety (free)
NVIDIA Nemotron 3.5 Content Safety is a compact 4B-parameter multimodal guardrail model from NVIDIA, fine-tuned from Google Gemma-3-4B. It moderates both inputs to and responses from LLMs and VLMs, accepting...
ctx 128K in Free out Free
NVIDIAFreeNemotron 3 Nano Omni (free)
NVIDIA Nemotron™ 3 Nano Omni is a 30B-A3B open multimodal model designed to function as a perception and context sub-agent in enterprise agent systems. It accepts text, image, video, and...
ctx 256K in Free out Free
NVIDIANemotron 3 Super
NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. Built on a hybrid Mamba-Transformer...
ctx 1M in $0.088 out $0.412