NVIDIA
availableShows if the model has enough results for an index.Nemotron 3 Ultra
Nemotron 3 Ultra is a reasoning model from NVIDIA. 25 benchmarks count toward its score, in 5 categories.
IndexOverall score out of 100.53.4 ±5.7
CoverageShare of the index weight with results.75%
SpeedOutput tokens per second.—
Input / 1MUS dollars per 1M input tokens.Free
Output / 1MUS dollars per 1M output tokens.Free
ContextMaximum tokens in one request.1M
EloLMArena rating and rank.N/A
The index is a score out of 100. The ± range shows how much it can change. A free tier is available.
CapabilitiesScore per category, out of 100.
Out of 100Results
25 counted| BenchmarkThe test name. | CategoryThe capability that the test measures. | ResultThe score from the publisher. | IndexThis result as a score out of 100. | RunThe settings of the run. | DateDate of the result. | Published byThe source of the result. |
|---|---|---|---|---|---|---|
| PinchBench | Agentic | 90.0% | — | — | — | Kilo Code |
| LiveCodeBench v6 | Coding | 89.0% | 58.0 | — | — | LiveCodeBench maintainers |
| Graduate-Level Google-Proof Q&A | Knowledge | 87.0% | 58.6 | — | — | David Rein et al. |
| GPQA Diamond | Knowledge | 87.0% | 58.6 | — | — | David Rein et al. |
| Massive Multitask Language Understanding Professional | Knowledge | 86.8% | 57.3 | — | — | Yubo Wang et al. |
| Artificial Analysis GPQA Diamond | Knowledge | 86.7% | 57.4 | — | — | Artificial Analysis |
| τ²-Bench Tool-Agent-User Evaluation | Agentic | 83.3% | 58.6 | — | — | Victor Barres et al. |
| Artificial Analysis Openness Index | Knowledge | 83.3% | — | — | — | Artificial Analysis |
| MMLU-ProX | Multilingual | 83.0% | — | — | — | MMLU-ProX authors |
| Artificial Analysis Harvey LAB-AA | Agentic | 81.7% | 61.2 | — | — | Artificial Analysis |
| Instruction Following Benchmark | Instruction | 81.7% | 61.1 | — | — | Benchmark authors |
| Artificial Analysis IFBench | Instruction | 81.4% | 71.7 | — | — | Artificial Analysis |
| Software Engineering Benchmark Verified | Coding | 71.9% | 55.2 | — | — | Carlos E. Jimenez et al. |
| τ³-Bench Tool-Agent-User Evaluation | Agentic | 70.9% | 55.4 | — | — | Sierra Research |
| Artificial Analysis Long Context Reasoning | Reasoning | 67.0% | 54.6 | — | — | Artificial Analysis |
| LongBench v2 | Reasoning | 61.9% | — | — | — | LongBench v2 authors |
| Artificial Analysis Coding Index | Coding | 49.3% | 53.7 | — | — | Artificial Analysis |
| Scientific Code Benchmark | Coding | 44.6% | 54.8 | — | — | Benchmark authors |
| BrowseComp | Agentic | 44.4% | 37.3 | — | — | OpenAI |
| Artificial Analysis SciCode | Coding | 40.3% | 48.8 | — | — | Artificial Analysis |
| Humanity's Last Exam with tools | Agentic | 37.4% | 50.9 | — | — | DeepSeek-AI |
| Terminal-Bench Hard | Agentic | 36.4% | — | — | — | Artificial Analysis |
| GDPval-AA normalized | Agentic | 33.1% | 61.0 | — | — | Artificial Analysis |
| Artificial Analysis EnterpriseOps-Gym | Agentic | 28.9% | 47.0 | — | — | Artificial Analysis |
| Artificial Analysis Humanity's Last Exam | Knowledge | 28.4% | 55.4 | — | — | Artificial Analysis |
| Humanity's Last Exam | Knowledge | 26.7% | 51.4 | — | — | Center for AI Safety et al. |
| Humanity's Last Exam without tools | Knowledge | 26.7% | 51.4 | — | — | OpenAI |
| Artificial Analysis Intelligence Index | Knowledge | 22.9% | 51.1 | — | — | Artificial Analysis |
| Artificial Analysis Agentic Index | Agentic | 21.7% | 54.8 | — | — | Artificial Analysis |
| Artificial Analysis Omniscience Accuracy | Knowledge | 21.6% | 40.6 | — | — | Artificial Analysis |
| Artificial Analysis AnalystAgent | Agentic | 6.3% | 45.9 | — | — | Artificial Analysis |
| Critical Physics Tasks | Reasoning | 3.1% | 44.9 | — | — | Artificial Analysis |
25 benchmarks count, from 27 of 32 results. A grey row does not count. Too few models took that benchmark.