Requesty
NVIDIA

nemotron-3-ultra-550b-a55b

NVIDIA Nemotron 3 Ultra is an open frontier reasoning and orchestration model with 55B active parameters out of 550B total (hybrid Transformer-Mamba MoE). Text input/output with up to 1M context, built for long-running agentic workflows: agent orchestration, coding agents, deep research, and complex enterprise tasks. Strong at multi-step reasoning and planning with high-throughput inference. Part of the NVIDIA Nemotron family.

ReasoningTool calling
Try nemotron-3-ultra-550b-a55bCompareDocs

Specifications

Context window1.0M tokens
Max output66K tokens
API typechat
AddedJun 9, 2026
Model IDnvidia/nemotron-3-ultra-550b-a55b
Data retentionYes
Used for trainingYes
Provider locationπŸ‡ΊπŸ‡Έ US
Privacy policyNVIDIA Privacy Policy→

Benchmarks

Released 2026-06-04
Coding Indexcoding
37.6%

Artificial Analysis Coding Index β€” a composite of coding evaluations including LiveCodeBench, SciCode and Terminal-Bench.

GPQA Diamondreasoning
86.7%

Graduate-level physics, chemistry & biology questions designed to resist Googling.

Intelligence Indexreasoning
47.7%

Artificial Analysis Intelligence Index β€” a composite of multiple evaluations measuring overall model capability.

Scores are sourced from official model cards, Artificial Analysis, and public leaderboards. Benchmarks measure specific skills and do not capture every aspect of model quality β€” always test on your own workload.

Pricing

Input / 1M
Free
Output / 1M
Free
Cache write
β€”
Cache read
β€”
Estimated cost
100K input + 10K outputFree
1M input + 100K outputFree
10M input + 1M outputFree

Requesty charges exactly what the upstream provider charges β€” no markup, no per-request fees. Prompt caching and smart routing can reduce effective cost by 30-80%.

Quickstart

Drop-in compatible with the OpenAI SDK. Change the base URL, swap in your Requesty API key, and set the model to nvidia/nemotron-3-ultra-550b-a55b.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
from openai import OpenAI

client = OpenAI(
    api_key="YOUR_REQUESTY_API_KEY",
    base_url="https://router.requesty.ai/v1",
)

response = client.chat.completions.create(
    model="nvidia/nemotron-3-ultra-550b-a55b",
    messages=[
        {"role": "user", "content": "Explain quantum computing in one paragraph."},
    ],
)

print(response.choices[0].message.content)

Other NVIDIA models

nemotron-3-nano-30b-a3b
262K ctxFree/1M
Coding 19
nemotron-3-super-120b-a12b
1.0M ctxFree/1M
Coding 31

Frequently asked questions

How much does nemotron-3-ultra-550b-a55b cost?
nemotron-3-ultra-550b-a55b is priced at Free per million input tokens and Free per million output tokens when accessed via Requesty. Requesty charges exactly what the upstream provider charges β€” we don't add markup.
What is the context window of nemotron-3-ultra-550b-a55b?
nemotron-3-ultra-550b-a55b has a context window of 1.0M tokens, with a maximum output of 66K tokens per response. That's roughly 1,398 words of input you can fit in a single prompt.
How does nemotron-3-ultra-550b-a55b perform on benchmarks?
nemotron-3-ultra-550b-a55b scores 86.7% on GPQA Diamond, 83.3% on τ²-Bench, 47.7% on Intelligence Index. See the full benchmark chart above for results across MMLU Pro, GPQA Diamond, SWE-Bench Verified, HumanEval, MATH, AIME, MMMU, and LiveBench.
What can nemotron-3-ultra-550b-a55b do?
nemotron-3-ultra-550b-a55b supports tool calling, extended reasoning. You can call it through any OpenAI-compatible client by pointing base_url to Requesty.
How do I use nemotron-3-ultra-550b-a55b with the OpenAI SDK?
Install the OpenAI SDK, set base_url to "https://router.requesty.ai/v1", set your API key to your Requesty key, and set the model to "nvidia/nemotron-3-ultra-550b-a55b". The Quickstart above shows Python, JavaScript and cURL snippets.

Access nemotron-3-ultra-550b-a55b through Requesty

One API key, 400+ models, OpenAI-compatible. No markup on provider prices, automatic failover, and smart caching built-in.

Get started freeAll NVIDIA models