DeepSeek V3.2 vs Nemotron 3 Ultra
DeepSeek V3.2 and Nemotron 3 Ultra tie on coding at 30.0 on the Vibe Coding Index, and DeepSeek V3.2 costs 2.7x less per output token ($0.88 vs $2.40 per million). Updated October 22, 2026; 55 benchmarks have a result for both.
DeepSeek V3.2 or Nemotron 3 Ultra: which to pick
DeepSeek V3.2 and Nemotron 3 Ultra are level for coding, 30.0 against 30.0 on the Vibe Coding Index, so price and context decide.
Pick Nemotron 3 Ultra for a larger context window (262K against 164K tokens) and work measured by AA Coding Index or LiveCodeBench (Vals AI), where it leads. Pick DeepSeek V3.2 for output tokens 2.7x cheaper ($0.88 vs $2.40 per million) and work measured by Epoch Capabilities Index or τ²-Bench Telecom, where it leads.
- Shared results
- 55
- DeepSeek V3.2 only
- 21
- Nemotron 3 Ultra only
- 16
- Like-for-like categories
- 8 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Level: 30.0 against 30.0, inside 2 points.
3 benchmarks behind each score
Intelligence
AA Intelligence Index
Nemotron 3 Ultra
Nemotron 3 Ultra leads, 22.9 against 21.5.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Nemotron 3 Ultra has a AA Agentic Index result, 20.1.
Directional only
Reasoning
GPQA Diamond
Nemotron 3 Ultra
Nemotron 3 Ultra leads, 86.1% against 80.3%.
Same board, both measured
Math
AIME
No call
Only DeepSeek V3.2 has a AIME result, 84.6%.
Directional only
Knowledge
MMLU Pro
Nemotron 3 Ultra
Nemotron 3 Ultra leads, 85.8% against 84.9%.
Same board, both measured
Web design
Design Arena: websites
DeepSeek V3.2
DeepSeek V3.2 leads, 1175 against 1145.
Same board, both measured
Long documents
Listed context window
Nemotron 3 Ultra
Nemotron 3 Ultra takes 262K tokens against 164K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
DeepSeek V3.2
DeepSeek V3.2 $0.0007, Nemotron 3 Ultra $0.0018 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only DeepSeek V3.2 has the number, $0.99.
Directional only
At a glance
- Coding index
- DeepSeek V3.2 30.0 from 3 benchmarks, Nemotron 3 Ultra 30.0 from 3 benchmarks.
- Widest category
- Knowledge, 5.1 points apart on average over 8 shared benchmarks.
- Widest single gap
- IFBench: DeepSeek V3.2 60.7% against Nemotron 3 Ultra 81.4%, 20.7 points apart on a 0 to 100 scale.
- Output price
- DeepSeek V3.2 $0.88 against Nemotron 3 Ultra $2.40 per million output tokens, 2.7x apart.
- Context window
- DeepSeek V3.2 takes 164K tokens, Nemotron 3 Ultra 262K.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | DeepSeek V3.2 | Bars | Nemotron 3 Ultra | Ties |
|---|---|---|---|---|
| Coding | 0 | 7 | ||
| Intelligence | 1 | 1 | ||
| Agentic work | 1 | 1 | ||
| Reasoning | 0 | 6 | ||
| Math | 0 | 2 | ||
| Knowledge | 0 | 8 | ||
| Web design | 5 | 3 | ||
| General | 5 | 15 | ||
| All shared boards | 12 | 43 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index07
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| SWE-bench Verified | 70.0% | Not measured | Only DeepSeek V3.2 measured |
| SWE-bench Multilingual | 59.0% | Not measured | Only DeepSeek V3.2 measured |
| LMArena WebDev | 1362 | Not measured | Only DeepSeek V3.2 measured |
| AA Coding Index | 44.2 | 49.3 | Nemotron 3 Ultra |
| LiveCodeBench (Vals AI) | 80.7% | 86.0% | Nemotron 3 Ultra |
| SWE-bench Verified (Vals AI) | 67.6% | 69.0% | Nemotron 3 Ultra |
| Vibe Code Bench (Vals AI) | 5.1% | 7.6% | Nemotron 3 Ultra |
| Code migration (Vals AI) | Not measured | 4.9% | Only Nemotron 3 Ultra measured |
| ProgramBench (Vals AI) | Not measured | 0.0% | Only Nemotron 3 Ultra measured |
| Terminal-Bench 2.1 (Vals AI) | Not measured | 50.9% | Only Nemotron 3 Ultra measured |
| LiveCodeBench (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| SciCode | Not measured | 40.3% | Only Nemotron 3 Ultra measured |
| Terminal-Bench Hard (Artificial Analysis run) | 35.6% | 36.4% | Nemotron 3 Ultra |
| Terminal-Bench | 39.6% | Not measured | Only DeepSeek V3.2 measured |
| FrontierCode | Not measured | 13.6% | Only Nemotron 3 Ultra measured |
| WeirdML | Not measured | 43.5% | Only Nemotron 3 Ultra measured |
| LMArena Text: coding | 1453 | 1468 | Nemotron 3 Ultra |
| LMArena Text: software and IT | 1449 | 1468 | Nemotron 3 Ultra |
| LMArena WebDev: React | 1350 | Not measured | Only DeepSeek V3.2 measured |
| LMArena WebDev: HTML | 1371 | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 23 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| AA Intelligence Index | 21.5 | 22.9 | Nemotron 3 Ultra |
| Epoch Capabilities Index | 146.1 | 146.0 | DeepSeek V3.2 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| AA Agentic Index | Not measured | 20.1 | Only Nemotron 3 Ultra measured |
| Finance Agent (Vals AI) | Not measured | 37.7% | Only Nemotron 3 Ultra measured |
| Legal Research Bench (Vals AI) | Not measured | 15.4% | Only Nemotron 3 Ultra measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 31.7% | Only Nemotron 3 Ultra measured |
| τ²-Bench Telecom | 90.6% | 83.3% | DeepSeek V3.2 |
| APEX-Agents | 21.3% | 22.7% | Nemotron 3 Ultra |
| Vending-Bench 2 | $1,034 | Not measured | Only DeepSeek V3.2 measured |
| MCP Atlas | Not measured | 63.1% | Only Nemotron 3 Ultra measured |
| Neither model has a result on the other 34 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| GPQA Diamond | 80.3% | 86.1% | Nemotron 3 Ultra |
| GPQA Diamond (Artificial Analysis run) | 84.0% | 86.7% | Nemotron 3 Ultra |
| Humanity's Last Exam (Artificial Analysis run) | 24.6% | 28.4% | Nemotron 3 Ultra |
| GPQA Diamond (Epoch AI run) | Not measured | 85.4% | Only Nemotron 3 Ultra measured |
| ARC-AGI | 57.0% | Not measured | Only DeepSeek V3.2 measured |
| ARC-AGI-2 | 4.0% | Not measured | Only DeepSeek V3.2 measured |
| Mystery game puzzles | Not measured | 20.0% | Only Nemotron 3 Ultra measured |
| LMArena Text: hard prompts | 1434 | 1452 | Nemotron 3 Ultra |
| LMArena Text: expert | 1436 | 1472 | Nemotron 3 Ultra |
| LMArena Text: hard prompts in English | 1446 | 1453 | Nemotron 3 Ultra |
| Neither model has a result on the other 13 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| AIME | 84.6% | Not measured | Only DeepSeek V3.2 measured |
| MGSM (Vals AI) | 86.0% | Not measured | Only DeepSeek V3.2 measured |
| AA Math Index | 92.0 | Not measured | Only DeepSeek V3.2 measured |
| AIME 2025 (Artificial Analysis run) | 92.0% | Not measured | Only DeepSeek V3.2 measured |
| OTIS Mock AIME | Not measured | 86.7% | Only Nemotron 3 Ultra measured |
| ProofBench | Not measured | 2.0% | Only Nemotron 3 Ultra measured |
| LMArena Text: math | 1435 | 1457 | Nemotron 3 Ultra |
| LMArena Text: mathematical work | 1440 | 1461 | Nemotron 3 Ultra |
| MathArena | 16.0% | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 5 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index08
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| MMLU Pro | 84.9% | 85.8% | Nemotron 3 Ultra |
| LegalBench (Vals AI) | 76.1% | 82.1% | Nemotron 3 Ultra |
| CaseLaw (Vals AI) | 55.4% | Not measured | Only DeepSeek V3.2 measured |
| CorpFin (Vals AI) | 51.0% | 65.5% | Nemotron 3 Ultra |
| TaxEval (Vals AI) | 68.2% | 73.1% | Nemotron 3 Ultra |
| MedQA (Vals AI) | 93.9% | Not measured | Only DeepSeek V3.2 measured |
| MedCode (Vals AI) | Not measured | 38.6% | Only Nemotron 3 Ultra measured |
| MMLU Pro (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: science | 1446 | 1465 | Nemotron 3 Ultra |
| LMArena Text: medicine and healthcare | 1440 | 1456 | Nemotron 3 Ultra |
| LMArena Text: legal and government | 1425 | 1452 | Nemotron 3 Ultra |
| LMArena Text: business and finance | 1410 | 1434 | Nemotron 3 Ultra |
| Neither model has a result on the other 8 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index53
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| Design Arena: websites | 1175 | 1145 | DeepSeek V3.2 |
| Design Arena: UI components | 1146 | 1135 | DeepSeek V3.2 |
| Design Arena: game dev | 1131 | 1138 | Nemotron 3 Ultra |
| Design Arena: data viz | 1159 | 1141 | DeepSeek V3.2 |
| Design Arena: 3D | 1132 | 1131 | DeepSeek V3.2 |
| Design Arena: all code categories | 1163 | 1146 | DeepSeek V3.2 |
| Design Arena: SVG | 1031 | 1069 | Nemotron 3 Ultra |
| Design Arena: ASCII art | 1082 | 1100 | Nemotron 3 Ultra |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index515
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| IFBench | 60.7% | 81.4% | Nemotron 3 Ultra |
| Long context reasoning (AA-LCR) | 73.3% | 79.3% | Nemotron 3 Ultra |
| DTBench | 85.6% | 90.1% | Nemotron 3 Ultra |
| LMCA | 28.8% | 36.9% | Nemotron 3 Ultra |
| CL-bench | 12.4% | Not measured | Only DeepSeek V3.2 measured |
| CL-bench Life | 7.4% | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text | 1425 | 1445 | Nemotron 3 Ultra |
| LMArena Text: instruction following | 1413 | 1418 | Nemotron 3 Ultra |
| LMArena Text: creative writing | 1400 | 1399 | DeepSeek V3.2 |
| LMArena Text: English | 1439 | 1455 | Nemotron 3 Ultra |
| LMArena Text: non-English | 1409 | 1427 | Nemotron 3 Ultra |
| LMArena Text: Chinese | 1455 | 1497 | Nemotron 3 Ultra |
| LMArena Text: German | 1413 | 1472 | Nemotron 3 Ultra |
| LMArena Text: Japanese | 1374 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: Korean | 1370 | 1384 | Nemotron 3 Ultra |
| LMArena Text: French | 1429 | 1472 | Nemotron 3 Ultra |
| LMArena Text: Spanish | 1424 | 1455 | Nemotron 3 Ultra |
| LMArena Text: Russian | 1424 | 1417 | DeepSeek V3.2 |
| LMArena Text: Polish | 1419 | 1385 | DeepSeek V3.2 |
| LMArena Text: multi-turn | 1427 | 1411 | DeepSeek V3.2 |
| LMArena Text: longer queries | 1428 | 1437 | Nemotron 3 Ultra |
| LMArena Text: writing and language | 1407 | 1406 | DeepSeek V3.2 |
| LMArena Text: entertainment and media | 1395 | 1397 | Nemotron 3 Ultra |
| Neither model has a result on the other 25 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | DeepSeek V3.2 | Nemotron 3 Ultra | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
DeepSeek V3.2 costs less
Repository review
50K input + 3K output tokens
DeepSeek V3.2 costs less
Agent loop
200K input + 10K output tokens
DeepSeek V3.2 costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | DeepSeek V3.2 | NVIDIA Nemotron 3 Ultra |
|---|---|---|
| Maker | DeepSeek | NVIDIA |
| Weights | Open weights | Open weights |
| Released | December 1, 2025 | June 4, 2026 |
| Context window | 164K | 262K |
| Input $/Mtok | $0.29 | $0.60 |
| Output $/Mtok | $0.88 | $2.40 |
| Coding index | 30.0 | 30.0 |
| Independent measurements behind it | 3 | 3 |
| Cost per solved task | $0.99 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is DeepSeek V3.2 better than Nemotron 3 Ultra for coding?
DeepSeek V3.2 and Nemotron 3 Ultra are level for coding, 30.0 against 30.0 on the Vibe Coding Index, so price and context decide. Pick Nemotron 3 Ultra for a larger context window (262K against 164K tokens) and work measured by AA Coding Index or LiveCodeBench (Vals AI), where it leads. Pick DeepSeek V3.2 for output tokens 2.7x cheaper ($0.88 vs $2.40 per million) and work measured by Epoch Capabilities Index or τ²-Bench Telecom, where it leads.
Which is cheaper, DeepSeek V3.2 or Nemotron 3 Ultra?
DeepSeek V3.2 costs $0.29 per million input tokens and $0.88 per million output tokens; Nemotron 3 Ultra costs $0.60 and $2.40. Cost per solved task is known for DeepSeek V3.2 only.
Which has the bigger context window, DeepSeek V3.2 or Nemotron 3 Ultra?
Nemotron 3 Ultra, with 262K tokens against 164K.
How many benchmarks is the DeepSeek V3.2 vs Nemotron 3 Ultra comparison based on?
55 benchmarks have a result for both models as of October 22, 2026. DeepSeek V3.2's coding index rests on 3 independent benchmarks and Nemotron 3 Ultra's on 3. Where only one model has a result, the page says so instead of naming a winner.