Model comparison
Nemotron 3 Ultra vs Phi-4
Nemotron 3 Ultra is the stronger model overall, scoring 42.5 to 31.2 on the Noometry Index. Phi-4 costs 11× less per token, which makes it the better buy when Nemotron 3 Ultra's lead doesn't matter for your workload.
Last verified . 20 shared benchmarks.
NVIDIA
MicrosoftSummary
- They share 20 benchmarks with published results for both. Nemotron 3 Ultra scores higher in 9 categories and Phi-4 in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Nemotron 3 Ultra leads 66.3 to 40.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 86.7% for Nemotron 3 Ultra and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.50 / $2.20 for Nemotron 3 Ultra.
- Nemotron 3 Ultra accepts more context: 262K tokens versus 128K.
Side by side
| Nemotron 3 Ultra | Phi-4 | |
|---|---|---|
| Provider | NVIDIA | Microsoft |
| Noometry Index | 42.5 | 31.2 |
| Released | 2026-06-04 | 2024-12-11 |
| Weights | Open | Open |
| Context window | 262K | 128K |
| Max output | 128K | 4K |
| Input $ / M tokens | $0.50 | $0.07 |
| Output $ / M tokens | $2.20 | $0.14 |
| Results tracked | 30 | 37 |
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), Phi-4: 34.4 (#239)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| LMArena Coding | 1468 | 1231 |
| FrontierCode | 13.6% | — |
| SciCode | 40.3% | — |
| WeirdML | 43.5% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
Agentic & Tool Use Too close to call
Nemotron 3 Ultra: 23.4 (#126), Phi-4: 22.8 (#128)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| APEX-Agents | 22.7% | — |
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning Nemotron 3 Ultra leads
Nemotron 3 Ultra: 31.0 (#82), Phi-4: 17.7 (#291)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| Chess Puzzles | 12% | 1% |
| LMArena Hard Prompts | 1452 | 1220 |
| Epoch Capabilities Index | 146.17 | 130.42 |
| CritPt | 3.1% | — |
| LiveBench Reasoning | — | 47.8% |
| Mystery Game Puzzles | 20% | — |
| DTBench | 90.1% | — |
| LiveBench Data Analysis | — | 45.2% |
| LMCA | 36.9% | — |
| LiveBench | — | 41.6% |
Math Nemotron 3 Ultra leads
Nemotron 3 Ultra: 35.3 (#187), Phi-4: 20.8 (#285)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 86.7% | 13.8% |
| LMArena Math | 1457 | 1246 |
| ProofBench | 2% | — |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
Knowledge Nemotron 3 Ultra leads
Nemotron 3 Ultra: 52.5 (#61), Phi-4: 32.6 (#209)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| GPQA Diamond | 85.4% | 56.1% |
| LMArena Expert | 1472 | 1203 |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| MMLU | — | 84.8% |
Multilingual Nemotron 3 Ultra leads
Nemotron 3 Ultra: 53.5 (#64), Phi-4: 37.2 (#237)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| LMArena Non-English | 1427 | 1197 |
| LMArena Chinese | 1497 | 1212 |
| LMArena French | 1472 | 1224 |
| LMArena German | 1471 | 1222 |
| LMArena Korean | 1386 | 1151 |
| LMArena Russian | 1417 | 1209 |
| LMArena Spanish | 1454 | 1234 |
| LMArena Japanese | — | 1158 |
Instruction Following Nemotron 3 Ultra leads
Nemotron 3 Ultra: 74.7 (#88), Phi-4: 60.4 (#251)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1418 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context Nemotron 3 Ultra leads
Nemotron 3 Ultra: 43.9 (#81), Phi-4: 36.9 (#226)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1437 | 1217 |
Writing & Preference Nemotron 3 Ultra leads
Nemotron 3 Ultra: 66.3 (#36), Phi-4: 40.5 (#244)
| Benchmark | Nemotron 3 Ultra | Phi-4 |
|---|---|---|
| LMArena Text | 1445 | 1217 |
| LMArena Creative Writing | 1398 | 1182 |
| LMArena Multi-Turn | 1411 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench Creative Writing | 1692 | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Nemotron 3 Ultra better than Phi-4?
Nemotron 3 Ultra is the stronger model overall, scoring 42.5 to 31.2 on the Noometry Index. Phi-4 costs 11× less per token, which makes it the better buy when Nemotron 3 Ultra's lead doesn't matter for your workload.
Which is cheaper, Nemotron 3 Ultra or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; Nemotron 3 Ultra lists at $0.50 and $2.20.
Is Nemotron 3 Ultra or Phi-4 better for coding?
Nemotron 3 Ultra scores higher on coding benchmarks: 38.1 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
Nemotron 3 Ultra does, with 262K tokens against 128K.
How many benchmarks do Nemotron 3 Ultra and Phi-4 share?
20 benchmarks have published results for both models. Nemotron 3 Ultra has 30 scored results on Noometry and Phi-4 has 37.