Noometry

NVIDIA, open weights

Nemotron 3 Ultra

Nemotron 3 Ultra by NVIDIA ranks 113th of 354 ranked models on the Noometry Index as of October 2026, with a score of 42.5. Its strongest category is writing & preference, where it ranks 36th. API pricing starts at $0.50 per million input tokens and $2.20 per million output tokens, with a 262K-token context window.

Last verified

Noometry rank
#113 of 354
Index score
42.5
Evidence
Confirmed 30 results
Provider
NVIDIA
Released
June 4, 2026
Weights
Open weights
Reasoning
Yes
Context window
262K
Max output
128K
Input price
$0.50 / M
Output price
$2.20 / M
Blended price
$0.93 / M
Output speed
Not measured
Value
#104 of 219
Knowledge cutoff
Unknown
Input
text
Hugging Face
nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16

Category scores

Each category score combines every public result we have in that category.

  1. Coding 38.1
  2. Agentic & Tool Use 23.4
  3. Reasoning 31.0
  4. Math 35.3
  5. Knowledge 52.5
  6. Multilingual 53.5
  7. Instruction Following 74.7
  8. Long Context 43.9
  9. Writing & Preference 66.3
CategoryScoreRankResults
Coding38.1#1824
Agentic & Tool Use23.4#1261
Reasoning31.0#826
Math35.3#1873
Knowledge52.5#612
Multilingual53.5#641
Instruction Following74.7#881
Long Context43.9#811
Writing & Preference66.3#364

Strengths and weaknesses

Categories where Nemotron 3 Ultra places highest and lowest among the models ranked in each, with its score against that category's median.

Strongest categories

CategoryScorevs medianRank
Writing & Preference66.3+12.5#36 of 312, top 12%
Knowledge52.5+15.2#61 of 314, top 20%
Multilingual53.5+6.1#64 of 297, top 22%

Weakest categories

CategoryScorevs medianRank
Agentic & Tool Use23.4−6.9#126 of 154, top 82%
Math35.3−1.3#187 of 327, top 58%
Coding38.1−0.6#182 of 340, top 54%

Closest competitors

The models ranked just above and below Nemotron 3 Ultra. When scores are this close, price and speed are often the better way to choose.

ModelRankScoreBlended $/MSpeed
GPT-5.3 Chat#10942.8$4.81—Compare
GPT-5.5 Instant#11042.7——Compare
GPT-5.2 Codex#11142.6$4.81—Compare
Qwen3.5-Flash#11242.5$0.18—Compare
Hunyuan T1 20250711#11442.5——Compare
DeepSeek-R1#11542.3$0.9110Compare
Step 3.5 Flash#11642.3$0.15—Compare
Qwen3.6 27B#11742.2$1.35—Compare

Sponsored placements are available on pages like this one. Advertise on Noometry

Benchmark results

Every published result we track, with its source. Bold rows are the ones used for ranking; where several exist we prefer independent runs over self-reported numbers.

Coding

BenchmarkScorePositionSettingSourceDate
FrontierCode13.6%#35 of 37, top 95%Epoch AI
SciCode40.3%#77 of 121, top 64%Epoch AI
WeirdML43.5%#65 of 119, top 55%Epoch AI
LMArena Coding1468#60 of 294, top 21%LMArena2026-10-08

Agentic & Tool Use

BenchmarkScorePositionSettingSourceDate
APEX-Agents22.7%#47 of 49, top 96%Epoch AI

Reasoning

BenchmarkScorePositionSettingSourceDate
CritPt3.1%#67 of 134, top 50%Epoch AI
Chess Puzzles12%#79 of 129, top 62%Epoch AI2026-08-10
LMArena Hard Prompts1452#57 of 297, top 20%LMArena2026-10-08
Mystery Game Puzzles20%#41 of 74, top 56%Epoch AI2026-08-27
DTBench90.1%#40 of 151, top 27%Epoch AI
LMCA36.9%#59 of 125, top 48%Epoch AI
Epoch Capabilities Index146.17#75 of 213, top 36%Epoch AI2026-06-04

Math

BenchmarkScorePositionSettingSourceDate
OTIS Mock AIME 2024-202586.7%#62 of 173, top 36%Epoch AI2026-08-10
ProofBench2%#74 of 77, top 97%Epoch AI
LMArena Math1457#55 of 285, top 20%LMArena2026-10-08

Knowledge

BenchmarkScorePositionSettingSourceDate
GPQA Diamond85.4%#63 of 186, top 34%Epoch AI2026-08-10
LMArena Expert1472#48 of 273, top 18%LMArena2026-10-08

Multilingual

BenchmarkScorePositionSettingSourceDate
LMArena Non-English1427#64 of 297, top 22%LMArena2026-10-08
LMArena Chinese1497#46 of 285, top 17%LMArena2026-10-08
LMArena French1472#34 of 223, top 16%LMArena2026-10-08
LMArena German1471#25 of 231, top 11%LMArena2026-10-08
LMArena Korean1386#69 of 213, top 33%LMArena2026-10-08
LMArena Russian1417#85 of 283, top 31%LMArena2026-10-08
LMArena Spanish1454#45 of 226, top 20%LMArena2026-10-08

Instruction Following

BenchmarkScorePositionSettingSourceDate
LMArena Instruction Following1418#77 of 298, top 26%LMArena2026-10-08

Long Context

BenchmarkScorePositionSettingSourceDate
LMArena Longer Query1437#69 of 291, top 24%LMArena2026-10-08

Writing & Preference

BenchmarkScorePositionSettingSourceDate
LMArena Text1445#54 of 297, top 19%LMArena2026-10-08
LMArena Creative Writing1398#84 of 295, top 29%LMArena2026-10-08
EQ-Bench Creative Writing1692#31 of 115, top 27%EQ-Bench
LMArena Multi-Turn1411#108 of 295, top 37%LMArena2026-10-08

API pricing by provider

RouteInput $/MOutput $/MCached input $/MChecked
fireworks$0.60$2.40$0.122026-10-10
openrouter$0.50$2.20$0.102026-10-10
together$0.60$3.60$0.202026-10-10

All NVIDIA API prices → Estimate your cost →

Compare Nemotron 3 Ultra

Other NVIDIA models

Frequently asked questions

How good is Nemotron 3 Ultra?

Nemotron 3 Ultra by NVIDIA ranks 113th of 354 ranked models on the Noometry Index as of October 2026, with a score of 42.5. Its strongest category is writing & preference, where it ranks 36th. API pricing starts at $0.50 per million input tokens and $2.20 per million output tokens, with a 262K-token context window.

How much does Nemotron 3 Ultra cost?

Nemotron 3 Ultra costs $0.50 per million input tokens and $2.20 per million output tokens on openrouter, with cached input at $0.10.

What is Nemotron 3 Ultra's context window?

Nemotron 3 Ultra accepts up to 262K tokens of input and can write up to 128K tokens in one response.

Is Nemotron 3 Ultra open source?

Yes. Nemotron 3 Ultra's weights are downloadable from Hugging Face (nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16); check the license for commercial terms.

What are Nemotron 3 Ultra's strengths and weaknesses?

Relative to other ranked models, Nemotron 3 Ultra places best in writing & preference, knowledge, multilingual and lowest in agentic & tool use, math, coding.

What is Nemotron 3 Ultra best at?

Its best category is writing & preference, where it ranks 36th on Noometry.