- AIME25 (reasoning on) 78.5
- MATH500 (reasoning on) 95.4
- GPQA (reasoning on) 53.2
- LCB (reasoning on, version not stated) 51.8
- BFCL v3 61.1
- IFEVAL-Prompt (reasoning on); Instruction 92 87.9
- Mean of Tau2 Airline 33.3 / Retail 39.8 / Telecom 33 (reasoning on) 35.4
- IFBench-Prompt (reasoning off); Instruction 44.2 43.2
- Orak (reasoning off)
- HaluEval (reasoning off) 62.2
- RULER (128k, reasoning off) 91.1
- EQ-Bench3 (reasoning off)
Nemotron 3 Nano 4B
Nemotron 3 Nano 4B.
13
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
NVIDIA Mar 2026 Nemotron 3 Nano 4B
Open weights Small
128.4
#285 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
- IFBench 43.2% #71 /80
- MultiChallenge 38.7% #37 /42
- AgentIF (2026-10 setting) 58.5% #18 /24
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| IFEval | 87.9% | #55 of 135 | Lab model card | |
| RULER | 91.1% | #19 of 42 | Lab model card | |
| AgentIF (2026-10 setting) | 58.5% | #18 of 24 | Official leaderboard | |
| MultiChallenge | 38.7% | #37 of 42 | Official leaderboard | |
| IFBench | 43.2% | #71 of 80 | Lab model card | |
| EQ-Bench 3 | 588 | #77 of 78 | Official leaderboard | |
| HaluEval | 62.2% | #1 of 1 | Lab model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MATH-500 | 95.4% | #39 of 70 | Lab model card | |
| AIME 2025 | 78.5% | #90 of 143 | Lab model card | |
| GPQA Diamond | 53.2% | #237 of 302 | Lab model card |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench | 51.8% | #28 of 54 | Lab model card |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Berkeley Function Calling Leaderboard (BFCL) | 61.1% | #35 of 41 | Lab model card | |
| tau2-bench | 35.4% | #77 of 79 | Lab model card |
Its lab's numbers put it at 128.4, number 285 of 404
The Self-Reported Index fits one score to the 9 capability results NVIDIA reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 125.5 to 131.4. How we compute it.
- MATH-500 95.4
- RULER 91.1
- IFEval 87.9
- AIME 2025 78.5
- Berkeley Function Calling Leaderboard (BFCL) 61.1
- GPQA Diamond 53.2
- LiveCodeBench 51.8
- IFBench 43.2
- tau2-bench 35.4
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.