# Nemotron 3 Ultra Benchmark Scores & Performance

> Nemotron 3 Ultra by NVIDIA scores 40.39/100 overall, ranking #171 out of 486 AI models.

Canonical page: https://benchlm.ai/models/nemotron-3-ultra

Last updated: September 15, 2026

## Model Details

| Property | Value |
|----------|-------|
| Creator | NVIDIA |
| Source Type | Open Weight |
| Reasoning Type | Reasoning |
| Context Window | 1M |
| Overall Score | 40.39/100 |
| Overall Rank | #171 of 486 |

## Family & Coverage

- Family: Nemotron 3 Ultra
- Variant: base
- Benchmarks covered: 43 of 439
- Coverage note: BenchLM currently has partial benchmark coverage for this model, so the overall score is conservative.

## Agentic Benchmarks

| Benchmark | Score |
|-----------|-------|
| [Terminal-Bench 2.0](/benchmarks/terminal-bench-2) | 56.4% |
| [PinchBench](/benchmarks/pinchbench) | 90.0% |
| [BrowseComp](/benchmarks/browsecomp) | 44.4% |
| [τ³-bench results](/benchmarks/tau3-bench) | 70.9% |
| [GDPval-AA](/benchmarks/gdpvalaanormalized) | 33.1% |
| [HLE w/ tools](/benchmarks/hlewithtools) | 37.4% |
| [τ²-bench results](/benchmarks/tau2-bench) | 83.3% |
| [AA Harvey LAB](/benchmarks/aaharveylab) | 81.7% |
| [terminalBenchHard](/benchmarks/terminal-bench-hard) | 36.4% |
| [Terminal-Bench 2.1 (Vals)](/benchmarks/valsterminalbench21) | 50.9% |
| [AA Agentic Index](/benchmarks/aaagenticindex) | 21.7% |
| [GDPval-AA](/benchmarks/gdpvalaa) | 1091 |
| [AA-AnalystAgent](/benchmarks/aaanalystagent) | 6.3% |

## Coding Benchmarks

| Benchmark | Score |
|-----------|-------|
| [SWE-bench Verified](/benchmarks/swe-bench-verified) | 71.9% |
| [SWE Multilingual](/benchmarks/swe-bench-multilingual) | 67.7% |
| [LiveCodeBench v6](/benchmarks/livecodebench-v6) | 89.0% |
| [SciCode](/benchmarks/scicode) | 44.6% |
| [Terminal-Bench 2.0](/benchmarks/terminal-bench-2) | 56.4% |
| [AA-SciCode](/benchmarks/aascicode) | 40.3% |
| [LiveCodeBench (Vals)](/benchmarks/valslivecodebench) | 86.0% |
| [SWE-bench (Vals)](/benchmarks/valsswebench) | 69.0% |
| [AA Coding Index](/benchmarks/aacodingindex) | 49.3% |

## Multimodal & Grounded Benchmarks

| Benchmark | Score |
|-----------|-------|
| [Design Arena Website](/benchmarks/designarenawebsite) | 1145 |

## Reasoning Benchmarks

| Benchmark | Score |
|-----------|-------|
| [AA-LCR](/benchmarks/lcr) | 67.0% |
| [CritPt](/benchmarks/critpt) | 3.1% |
| [LongBench v2](/benchmarks/longbench-v2) | 61.9% |

## Knowledge Benchmarks

| Benchmark | Score |
|-----------|-------|
| [GPQA](/benchmarks/gpqa) | 87% |
| [GPQA-D](/benchmarks/gpqa-diamond) | 87.0% |
| [HLE](/benchmarks/hle) | 26.7% |
| [HLE w/o tools](/benchmarks/hlenotools) | 26.7% |
| [MMLU-Pro](/benchmarks/mmlu-pro) | 86.8% |
| [AA-Omniscience Accuracy](/benchmarks/omniscienceaccuracy) | 21.6% |
| [Artificial Analysis Intelligence Index](/benchmarks/artificialanalysis) | 23.4% |
| [AA-GPQA Diamond](/benchmarks/aagpqadiamond) | 86.7% |
| [AA-HLE](/benchmarks/aahle) | 28.4% |
| [AA-Omniscience Index](/benchmarks/aaomniscienceindex) | -0.4% |
| [AA-Omniscience Hallucination Rate](/benchmarks/omnisciencehallucinationrate) | 29.7% |
| [GPQA Diamond (Vals)](/benchmarks/valsgpqadiamond) | 86.1% |
| [MMLU-Pro (Vals)](/benchmarks/valsmmlupro) | 85.8% |
| [AA Openness Index](/benchmarks/aaopennessindex) | 83.3% |

## Instruction Following Benchmarks

| Benchmark | Score |
|-----------|-------|
| [IFBench](/benchmarks/ifbench) | 81.7% |
| [AA-IFBench](/benchmarks/aaifbench) | 81.4% |

## Multilingual Benchmarks

| Benchmark | Score |
|-----------|-------|
| [MMLU-ProX](/benchmarks/mmluprox) | 83% |

## Other NVIDIA Models

- [Nemotron 3 Nano 30B](/models/nemotron-3-nano-30b) - Score: 50.88
- [Nemotron 3 Super 120B A12B](/models/nemotron-3-super-120b-a12b) - Score: 48.94
- [Nemotron 3 Super 100B](/models/nemotron-3-super-100b) - Score: 48.07
- [Nemotron Ultra 253B](/models/nemotron-ultra-253b) - Score: 42.51
- [Nemotron 3 Nano Omni 30B A3B](/models/nemotron-3-nano-omni-30b-a3b) - Score: 40.77
- [Nemotron 3.5 Lightning 30B A3B NVFP4](/models/nemotron-3-5-lightning-30b-a3b-nvfp4) - Score: 21.02
- [Nemotron-4 15B](/models/nemotron-4-15b) - Score: 15.92
- [Cosmos3-Edge](/models/cosmos3-edge) - Score: not computed
- [Audio Flamingo 3 7B](/models/audio-flamingo-3-7b) - Score: not computed
- [Parakeet CTC 1.1B](/models/parakeet-ctc-1-1b) - Score: not computed
