Model comparison
Nemotron 3 Ultra vs Nemotron 4 340b Instruct
Nemotron 3 Ultra is the stronger model overall, scoring 42.5 to 35.9 on the Noometry Index.
Last verified . 16 shared benchmarks.
Summary
- They share 16 benchmarks with published results for both. Nemotron 3 Ultra scores higher in 8 categories and Nemotron 4 340b Instruct in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Nemotron 3 Ultra leads 66.3 to 42.7.
Side by side
| Nemotron 3 Ultra | Nemotron 4 340b Instruct | |
|---|---|---|
| Provider | NVIDIA | NVIDIA |
| Noometry Index | 42.5 | 35.9 |
| Released | 2026-06-04 | 2024-06-14 |
| Weights | Open | Open |
| Context window | 262K | — |
| Max output | 128K | — |
| Input $ / M tokens | $0.50 | — |
| Output $ / M tokens | $2.20 | — |
| Results tracked | 30 | 17 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Nemotron 3 Ultra leads
Nemotron 3 Ultra: 38.1 (#182), Nemotron 4 340b Instruct: 35.2 (#228)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Coding | 1468 | 1209 |
| FrontierCode | 13.6% | — |
| SciCode | 40.3% | — |
| WeirdML | 43.5% | — |
Agentic & Tool Use Not comparable
Nemotron 3 Ultra: 23.4 (#126), Nemotron 4 340b Instruct: —
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| APEX-Agents | 22.7% | — |
Reasoning Nemotron 3 Ultra leads
Nemotron 3 Ultra: 31.0 (#82), Nemotron 4 340b Instruct: 23.4 (#178)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Hard Prompts | 1452 | 1202 |
| CritPt | 3.1% | — |
| Chess Puzzles | 12% | — |
| Mystery Game Puzzles | 20% | — |
| DTBench | 90.1% | — |
| LMCA | 36.9% | — |
| Epoch Capabilities Index | 146.17 | — |
Math Too close to call
Nemotron 3 Ultra: 35.3 (#187), Nemotron 4 340b Instruct: 34.3 (#195)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Math | 1457 | 1216 |
| OTIS Mock AIME 2024-2025 | 86.7% | — |
| ProofBench | 2% | — |
Knowledge Nemotron 3 Ultra leads
Nemotron 3 Ultra: 52.5 (#61), Nemotron 4 340b Instruct: 32.0 (#216)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Expert | 1472 | 1171 |
| GPQA Diamond | 85.4% | — |
Multilingual Nemotron 3 Ultra leads
Nemotron 3 Ultra: 53.5 (#64), Nemotron 4 340b Instruct: 37.9 (#233)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Non-English | 1427 | 1206 |
| LMArena Chinese | 1497 | 1214 |
| LMArena French | 1472 | 1197 |
| LMArena German | 1471 | 1192 |
| LMArena Korean | 1386 | 1148 |
| LMArena Russian | 1417 | 1219 |
| LMArena Spanish | 1454 | 1195 |
| LMArena Japanese | — | 1128 |
Instruction Following Nemotron 3 Ultra leads
Nemotron 3 Ultra: 74.7 (#88), Nemotron 4 340b Instruct: 63.1 (#232)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Instruction Following | 1418 | 1203 |
Long Context Nemotron 3 Ultra leads
Nemotron 3 Ultra: 43.9 (#81), Nemotron 4 340b Instruct: 37.1 (#221)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Longer Query | 1437 | 1225 |
Writing & Preference Nemotron 3 Ultra leads
Nemotron 3 Ultra: 66.3 (#36), Nemotron 4 340b Instruct: 42.7 (#233)
| Benchmark | Nemotron 3 Ultra | Nemotron 4 340b Instruct |
|---|---|---|
| LMArena Text | 1445 | 1225 |
| LMArena Creative Writing | 1398 | 1203 |
| LMArena Multi-Turn | 1411 | 1214 |
| EQ-Bench Creative Writing | 1692 | — |
Frequently asked questions
Is Nemotron 3 Ultra better than Nemotron 4 340b Instruct?
Nemotron 3 Ultra is the stronger model overall, scoring 42.5 to 35.9 on the Noometry Index.
Is Nemotron 3 Ultra or Nemotron 4 340b Instruct better for coding?
Nemotron 3 Ultra scores higher on coding benchmarks: 38.1 versus 35.2 in the Noometry coding category.
How many benchmarks do Nemotron 3 Ultra and Nemotron 4 340b Instruct share?
16 benchmarks have published results for both models. Nemotron 3 Ultra has 30 scored results on Noometry and Nemotron 4 340b Instruct has 17.