NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/Nemotron 3 Ultra
NVIDIA logo

Nemotron 3 Ultra

by NVIDIA · Released Jun 2026

Open Source1M Context
45.6
avg score
Rank #169
Compare
Better than 46% of all models
Context
1.0M tokens (~500 books)
Input $/1M
$0.50
Output $/1M
$2.20
Type
text
License
Open Source
Benchmarks
29 tested
Data as of
About

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...

Tested on 29 benchmarks · BenchGecko score 45.6. Top scores: Artificial Analysis · GPQA Diamond (86.7%), OTIS Mock AIME 2024-2025 (86.7%), Dtbench (83.5%).

Looking for similar performance at lower cost?
gpt-oss-20b scores 45.0 (99% as good) at $0.02/1M input · 96% cheaper
Capabilities
coding
43.8
#125 globally
reasoning
39.8
#101 globally
math
70.6
#41 globally
knowledge
46.6
#153 globally
speed
47.0
#38 globally
general
35.2
#102 globally
language
55.2
#117 globally
Benchmark Scores
Compare All
Tested on 29 benchmarks · Ranked across 7 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
LiveBench — Coding

Regularly refreshed coding problems that avoid data contamination. New problems added monthly to prevent memorization.

71.3·
LiveBench — Agentic Coding

LiveBench coding tasks that require multi-step reasoning and tool use. Tests planning and execution of complex coding workflows.

46.7·
WeirdML

Unusual and adversarial machine learning challenges. Tests robustness of reasoning about edge cases in ML systems.

43.5·
reasoningCompare reasoning →
LiveBench — Data Analysis

Fresh data analysis tasks testing ability to interpret tables, charts, and statistical data.

42.0·
LiveBench — Reasoning

Regularly refreshed reasoning problems testing logical deduction, spatial reasoning, and analytical thinking.

37.5·
mathCompare math →
OTIS Mock AIME 2024-2025

Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.

86.7·
LiveBench — Mathematics

Regularly updated math problems that test numerical reasoning, algebra, calculus, and combinatorics.

54.5·
Quick compare:Alibaba Qwen logovs Qwen3.8 27BGoogle DeepMind logovs Gemini 2.0 Flash LiteDeepSeek logovs R1
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · NVIDIA Nemotron 3
Dec 2025
$0.05/M in262Kctx
Dec 2025
$0.00/M in(-0.05)256Kctx(-6K)
Apr 2026
$0.00/M in256Kctx
Mar 2026
10.2
$0.08/M in(+0.08)262Kctx(+6K)18 benchmarks
Mar 2026
$0.00/M in(-0.08)262Kctx
Jun 2026
45.6+35.4
$0.50/M in(+0.50)1.0Mctx(+738K)29 benchmarks
Jun 2026
$0.60/M in(+0.10)512Kctx(-488K)
Jun 2026
$0.00/M in(-0.60)1.0Mctx(+488K)
Jun 2026
$0.20/M in(+0.20)131Kctx(-869K)
Jun 2026
$0.00/M in(-0.20)128Kctx(-3K)
Aug 2026
$0.06/M in(+0.06)262Kctx(+134K)7 benchmarks
Aug 2026
$0.00/M in(-0.06)1.0Mctx(+738K)
See the full Nemotron 3 family →
Similar Models
Alibaba Qwen logo
Qwen3.8 27B
Alibaba Qwen
45.6$0.42/1M
Google DeepMind logo
Gemini 2.0 Flash Lite
Google DeepMind
45.3$0.07/1M
DeepSeek logo
R1
DeepSeek
46.0$0.70/1M
Links
Info
NVIDIAPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@NVIDIA
Source Code
GitHubHugging Face
BenchGecko API
nemotron-3-ultra-550b-a55b
Specifications
  • Typetext
  • Context1.0M tokens (~500 books)
  • ReleasedJun 2026
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.003
Available On
NVIDIA logoNVIDIA$0.50
Categories
codingreasoningmathknowledgespeedgenerallanguage
Learn More
context-windowtransformeropen weightstokens
Share & Export
Tweet
Related Models
Alibaba Qwen logoQwen3.8 27B
45.6
Google DeepMind logoGemini 2.0 Flash Lite
45.3
DeepSeek logoR1
46.0
OpenAI logogpt-oss-20b
45.0
Nemotron 3 Ultra is an open-source text AI model by NVIDIA, released in June 2026. It has an average benchmark score of 45.6. Context window: 1M tokens.

Related Models

Qwen3.8 27B · Alibaba QwenGemini 2.0 Flash Lite · Google DeepMindR1 · DeepSeekgpt-oss-20b · OpenAIQwen3 235B A22B Instruct 2507 · Alibaba Qwen

Benchmarks

Artificial Analysis · GPQA DiamondOTIS Mock AIME 2024-2025DtbenchArtificial Analysis · tau2-Bench TelecomArtificial Analysis · IFBench

Related Pages

NVIDIA · ProviderNVIDIA · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-06-22

  • Nemotron 3 Ultra by NVIDIA. BenchGecko score 45.6, rank 169 of 312 scored models (normalized average of public benchmark scores).
  • List price $0.50 input · $2.20 output per 1M tokens (as of 2026-06-22).

How to cite · data as of 2026-06-22

Nemotron 3 Ultra · benchmarks, pricing and providers. BenchGecko, data as of 2026-06-22. https://benchgecko.ai/model/nemotron-3-ultra-550b-a55b

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP