NVIDIA · model
Nemotron 3.5 Super VL 120B A12B benchmark scores
As of 2026-10-29, Nemotron 3.5 Super VL 120B A12B (NVIDIA) has measured scores on 23 benchmarks and estimated scores on 117 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 88.1% | estimated ± 2.7 pp, high confidence |
| AA-HLE | 30.7% | measured |
| AA-SciCode | 47.8% | measured |
| CritPt | 10.9% | measured |
| GDPval-AA | 33.1% | measured |
| GPQA Diamond | 89.5% | measured |
| HLE | 29.2% | measured |
| IFBench | 70.9% | measured |
| AA-LCR | 68.6% | measured |
| LiveCodeBench v6 | 88.2% | measured |
| SciCode | 47.7% | estimated ± 3.1 pp, high confidence |
| Terminal-Bench 2.1 | 66.7% | measured |
| BrowseComp | 78.0% | measured |
| HLE w/ tools | 49.8% | estimated ± 5.3 pp, medium confidence |
| IFEval | 92.8% | estimated ± 3.6 pp, high confidence |
| LongBench v2 | 60.0% | estimated ± 5.8 pp, low confidence |
| VITA-Bench | 44.2% | estimated ± 6.7 pp, low confidence |
| τ²-bench results | 93.2% | estimated ± 11.5 pp, low confidence |
| AA Coding Index | 42.0% | estimated ± 5.3 pp, medium confidence |
| AA-MMMU-Pro | 71.9% | estimated ± 2.1 pp, high confidence |
| APEX-Agents | 25.4% | estimated ± 9.0 pp, low confidence |
| APEX-Agents-AA | 24.2% | estimated ± 4.2 pp, high confidence |
| BioMysteryBench (human-difficult) | 44.4% | estimated ± 6.8 pp, low confidence |
| DeepSearchQA | 84.1% | estimated ± 7.3 pp, low confidence |
| SWE-bench Verified | 71.1% | measured |
| AutomationBench | 27.6% | estimated ± 10.5 pp, low confidence |
| BFCL v4 | 74.9% | estimated ± 2.7 pp, medium confidence |
| CyberGym | 63.9% | estimated ± 6.0 pp, medium confidence |
| JobBench | 32.3% | estimated ± 8.7 pp, medium confidence |
| SWE-bench Pro | 51.8% | measured |
| τ³-bench results | 18.9% | measured |
| WideResearch | 76.8% | estimated ± 3.5 pp, high confidence |
| AA Agentic Index | 26.6% | estimated ± 2.9 pp, high confidence |
| MMLU-Pro | 85.3% | estimated ± 2.5 pp, medium confidence |
| AA-IFBench | 56.8% | estimated ± 7.4 pp, medium confidence |
| Artificial Analysis Intelligence Index | 13.6% | measured |
| GPQA | 89.4% | estimated ± 1.2 pp, high confidence |
| MMLU | 89.7% | estimated ± 0.7 pp, high confidence |
| AA-Omniscience Accuracy | 26.2% | measured |
| MMMLU | 89.7% | estimated ± 0.7 pp, medium confidence |
| MMMU | 82.4% | estimated ± 3.7 pp, high confidence |
| AA MMLU-Pro | 84.6% | estimated ± 0.6 pp, high confidence |
| Gert Labs | 59.3% | estimated ± 7.4 pp, medium confidence |
| AA EnterpriseOps-Gym | 41.8% | estimated ± 3.3 pp, medium confidence |
| AA Harvey LAB v1.0 | 90.0% | estimated ± 9.8 pp, low confidence |
| AA Tau3 Banking | 33.9% | estimated ± 3.9 pp, high confidence |
| AA Terminal-Bench 2.1 | 72.0% | estimated ± 6.8 pp, medium confidence |
| ARC-AGI-1 | 88.3% | estimated ± 2.7 pp, high confidence |
| ARC-AGI-2 | 53.8% | estimated ± 10.1 pp, low confidence |
| CursorBench 3.1 | 38.6% | estimated ± 3.4 pp, low confidence |
| CursorBench 3.2 | 44.1% | estimated ± 3.6 pp, medium confidence |
| Terminal-Bench 3.0 | 7.0% | estimated ± 9.1 pp, low confidence |
| FrontierCode 1.1 Main | 4.7% | estimated ± 2.7 pp, low confidence |
| FrontierSWE v2 | 0.5% | estimated ± 9.0 pp, low confidence |
| MazeBench | 0.4% | estimated ± 5.4 pp, low confidence |
| OSWorld-Verified | 70.6% | estimated ± 5.8 pp, medium confidence |
| ParseBench | 58.5% | estimated ± 8.0 pp, low confidence |
| Vals GPQA Diamond | 88.8% | estimated ± 2.0 pp, high confidence |
| Vals LiveCodeBench | 85.1% | estimated ± 2.0 pp, low confidence |
| Vals MMLU-Pro | 86.4% | estimated ± 1.6 pp, high confidence |
| Vals SWE-bench | 64.1% | estimated ± 6.8 pp, medium confidence |
| Terminal-Bench 2.1 (Vals AI) | 57.5% | estimated ± 7.4 pp, medium confidence |
| AA AutomationBench | 33.4% | estimated ± 14.8 pp, low confidence |
| CWE-Bench-AA | 44.2% | estimated ± 5.8 pp, medium confidence |
| AA Cyber Index | 37.6% | estimated ± 13.3 pp, low confidence |
| DeepsecBench-AA | 4.9% | estimated ± 8.6 pp, medium confidence |
| GDP.pdf | 11.1% | estimated ± 4.9 pp, medium confidence |
| AA ITBench | 44.9% | estimated ± 4.1 pp, high confidence |
| AA Terminal-Bench 4.0 | 6.7% | estimated ± 11.6 pp, low confidence |
| CursorBench 4.0 | 23.8% | estimated ± 3.5 pp, medium confidence |
| CWE-bench v1 | 42.4% | estimated ± 6.4 pp, medium confidence |
| DeepSWE | 51.9% | estimated ± 7.5 pp, low confidence |
| FrontierCode 2.0 | 18.4% | estimated ± 7.1 pp, low confidence |
| HLE w/o tools | 31.8% | measured |
| OSWorld 2.1 | 49.0% | estimated ± 6.0 pp, low confidence |
| PostTrainBench v1.1 | 20.9% | estimated ± 4.7 pp, medium confidence |
| SWE Multilingual | 67.7% | measured |
| Toolathlon-Verified | 59.4% | estimated ± 3.1 pp, high confidence |
| OSWorld 2.1 Offline | 32.0% | estimated ± 7.9 pp, low confidence |
| Vibe Code Bench | 0.8% | estimated ± 8.1 pp, medium confidence |
| ARC-AGI-3 | 0.0% | estimated ± 3.6 pp, high confidence |
| ARC-AGI-3 Provider Adapter | 2.8% | estimated ± 7.5 pp, low confidence |
| Hermes Bench | 62.0% | estimated ± 2.9 pp, high confidence |
| Hermes Index | 29.2% | estimated ± 8.7 pp, medium confidence |
| CharXiv | 80.8% | measured |
| CharXiv w/o tools | 77.6% | estimated ± 2.2 pp, medium confidence |
| ExploitGym | 0.0% | estimated ± 9.2 pp, low confidence |
| AI2D_TEST | 93.4% | estimated ± 0.9 pp, medium confidence |
| C-Eval | 93.0% | estimated ± 0.9 pp, medium confidence |
| Claw-Eval | 62.7% | estimated ± 7.6 pp, low confidence |
| CountBench | 91.7% | estimated ± 1.3 pp, medium confidence |
| DynaMath | 85.9% | estimated ± 0.8 pp, medium confidence |
| ERQA | 65.5% | estimated ± 4.9 pp, high confidence |
| INCLUDE | 83.0% | estimated ± 2.0 pp, low confidence |
| MathVision | 78.2% | measured |
| MCP Atlas | 73.2% | estimated ± 10.7 pp, low confidence |
| MLVU (M-Avg) | 82.5% | estimated ± 1.5 pp, high confidence |
| MMLU-ProX | 82.8% | estimated ± 1.3 pp, high confidence |
| MMLU-Redux | 94.3% | estimated ± 0.8 pp, medium confidence |
| MMMU-Pro | 78.4% | estimated ± 2.9 pp, high confidence |
| NL2Repo | 34.2% | estimated ± 2.8 pp, high confidence |
| OmniDocBench 1.5 | 87.2% | estimated ± 2.2 pp, high confidence |
| QwenClawBench | 87.1% | estimated ± 6.0 pp, low confidence |
| RealWorldQA | 83.8% | estimated ± 1.5 pp, high confidence |
| ScreenSpot Pro | 73.2% | estimated ± 7.1 pp, medium confidence |
| SimpleVQA | 62.6% | estimated ± 7.3 pp, medium confidence |
| SuperGPQA | 70.1% | estimated ± 6.9 pp, medium confidence |
| Toolathlon | 47.5% | estimated ± 4.2 pp, high confidence |
| V* | 85.6% | measured |
| Video-MME (w/o subtitle) | 81.4% | estimated ± 4.9 pp, medium confidence |
| Video-MME (with subtitle) | 82.4% | estimated ± 1.9 pp, high confidence |
| VideoMMMU | 85.4% | estimated ± 1.5 pp, medium confidence |
| We-Math | 69.3% | estimated ± 2.1 pp, medium confidence |
| HealthBench Hard | 29.4% | estimated ± 8.6 pp, low confidence |
| LiveCodeBench Pro | 77.0% | estimated ± 6.0 pp, low confidence |
| MedXpertQA (MM) | 75.3% | estimated ± 6.4 pp, low confidence |
| MedXpertQA (Text) | 52.4% | estimated ± 4.5 pp, medium confidence |
| MMLU-Pro (Arcee) | 86.7% | estimated ± 3.4 pp, low confidence |
| React Native Evals | 72.0% | estimated ± 2.3 pp, medium confidence |
| ResearchClawBench | 17.7% | estimated ± 2.2 pp, medium confidence |
| SWE-Rebench | 57.0% | estimated ± 3.6 pp, medium confidence |
| BioMysteryBench (human-solvable) | 84.0% | estimated ± 0.6 pp, low confidence |
| DRACO | 75.1% | estimated ± 3.1 pp, medium confidence |
| FrontierCode 1.1 Extended | 53.0% | estimated ± 3.0 pp, low confidence |
| HealthBench (raw) | 39.6% | estimated ± 6.0 pp, low confidence |
| HealthBench (length-adjusted) | 53.9% | estimated ± 3.3 pp, medium confidence |
| HealthBench Professional | 58.1% | estimated ± 4.5 pp, low confidence |
| HealthBench Professional (raw) | 50.8% | estimated ± 4.5 pp, medium confidence |
| HLE-Verified | 0.0% | estimated ± 6.8 pp, low confidence |
| LABBench2 | 75.8% | estimated ± 2.4 pp, low confidence |
| STEMBench | 2.4% | estimated ± 0.2 pp, high confidence |
| SimpleQA | 26.6% | estimated ± 4.3 pp, medium confidence |
| LiveCodeBench | 73.7% | estimated ± 14.6 pp, low confidence |
| Agents' Last Exam | 25.6% | estimated ± 13.0 pp, low confidence |
| OpenHarmony Bench | 51.1% | estimated ± 2.4 pp, high confidence |
| BabyVision | 29.6% | measured |
| MMVU | 71.0% | estimated ± 3.3 pp, medium confidence |
| MRCRv2 | 67.7% | estimated ± 11.3 pp, low confidence |
| MMLongBench-Doc | 62.8% | measured |
| PinchBench | 90.1% | estimated ± 3.6 pp, medium confidence |