Model comparison
DeepSeek-V3 vs Nemotron 3 Ultra
Nemotron 3 Ultra is the stronger model overall, scoring 44.7 to 39.2 on the Noometry Index. DeepSeek-V3 costs 2.6× less per token, which makes it the better buy when Nemotron 3 Ultra's lead doesn't matter for your workload.
Last verified . 25 shared benchmarks.
DeepSeek
NVIDIASummary
- They share 25 benchmarks with published results for both. DeepSeek-V3 scores higher in 1 category and Nemotron 3 Ultra in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Nemotron 3 Ultra leads 52.5 to 37.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 86.7% for Nemotron 3 Ultra.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $0.60 / $2.40 for Nemotron 3 Ultra.
- Nemotron 3 Ultra accepts more context: 262K tokens versus 164K.
Side by side
| DeepSeek-V3 | Nemotron 3 Ultra | |
|---|---|---|
| Provider | DeepSeek | NVIDIA |
| Noometry Index | 39.2 | 44.7 |
| Released | 2024-12-26 | 2026-06-04 |
| Weights | Open | Open |
| Context window | 164K | 262K |
| Max output | 164K | 128K |
| Input $ / M tokens | $0.24 | $0.60 |
| Output $ / M tokens | $0.90 | $2.40 |
| Results tracked | 60 | 29 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 41.9 (#99), Nemotron 3 Ultra: 37.7 (#174)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| SciCode | 35.8% | 40.3% |
| WeirdML | 36.1% | 43.5% |
| LMArena Coding | 1368 | 1468 |
| FrontierCode | — | 13.6% |
| Aider Polyglot | 55.1% | — |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Nemotron 3 Ultra: —
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| METR Time Horizons | 49.6% | — |
Reasoning Nemotron 3 Ultra leads
DeepSeek-V3: 20.0 (#226), Nemotron 3 Ultra: 30.3 (#78)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| CritPt | 0% | 3.1% |
| LMArena Hard Prompts | 1365 | 1452 |
| DTBench | 64.8% | 90.1% |
| LMCA | 15.5% | 36.9% |
| Epoch Capabilities Index | 135.95 | 146.2 |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| Chess Puzzles | — | 12% |
| LiveBench Reasoning | 65.8% | — |
| Mystery Game Puzzles | — | 20% |
| LiveBench Data Analysis | 60.9% | — |
| BIG-Bench Hard | 87.5% | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math Nemotron 3 Ultra leads
DeepSeek-V3: 30.4 (#208), Nemotron 3 Ultra: 34.4 (#160)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 86.7% |
| LMArena Math | 1373 | 1456 |
| ProofBench | — | 2% |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge Nemotron 3 Ultra leads
DeepSeek-V3: 37.5 (#147), Nemotron 3 Ultra: 52.5 (#56)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| GPQA Diamond | 67.6% | 85.4% |
| LMArena Expert | 1352 | 1472 |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multilingual Nemotron 3 Ultra leads
DeepSeek-V3: 48.8 (#138), Nemotron 3 Ultra: 53.7 (#61)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| LMArena Non-English | 1358 | 1427 |
| LMArena Chinese | 1390 | 1498 |
| LMArena French | 1386 | 1475 |
| LMArena German | 1374 | 1471 |
| LMArena Korean | 1321 | 1387 |
| LMArena Russian | 1372 | 1415 |
| LMArena Spanish | 1358 | 1452 |
| LMArena Japanese | 1333 | — |
Instruction Following Nemotron 3 Ultra leads
DeepSeek-V3: 73.0 (#125), Nemotron 3 Ultra: 74.9 (#84)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| LMArena Instruction Following | 1345 | 1417 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |
Long Context Nemotron 3 Ultra leads
DeepSeek-V3: 34.0 (#248), Nemotron 3 Ultra: 43.9 (#77)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| LMArena Longer Query | 1353 | 1437 |
| Fiction.LiveBench | 50% | — |
Writing & Preference Nemotron 3 Ultra leads
DeepSeek-V3: 57.6 (#124), Nemotron 3 Ultra: 66.3 (#35)
| Benchmark | DeepSeek-V3 | Nemotron 3 Ultra |
|---|---|---|
| LMArena Text | 1375 | 1444 |
| LMArena Creative Writing | 1364 | 1392 |
| EQ-Bench Creative Writing | 1472 | 1692 |
| LMArena Multi-Turn | 1389 | 1412 |
| Short-Story Creative Writing | 77% | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Nemotron 3 Ultra?
Nemotron 3 Ultra is the stronger model overall, scoring 44.7 to 39.2 on the Noometry Index. DeepSeek-V3 costs 2.6× less per token, which makes it the better buy when Nemotron 3 Ultra's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3 or Nemotron 3 Ultra?
DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Nemotron 3 Ultra lists at $0.60 and $2.40.
Is DeepSeek-V3 or Nemotron 3 Ultra better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 41.9 versus 37.7 in the Noometry coding category.
Which has the bigger context window?
Nemotron 3 Ultra does, with 262K tokens against 164K.
How many benchmarks do DeepSeek-V3 and Nemotron 3 Ultra share?
25 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Nemotron 3 Ultra has 29.