Nemotron 3 Ultra vs Qwen3 235B A22B Instruct 2507
Nemotron 3 Ultra leads Qwen3 235B A22B Instruct 2507 on the Artificial Analysis Intelligence Index, 22.9 to 12.0, and Qwen3 235B A22B Instruct 2507 costs 3.1x less per output token ($0.775 vs $2.40 per million). Updated October 23, 2026; 17 benchmarks have a result for both.
Nemotron 3 Ultra or Qwen3 235B A22B Instruct 2507: which to pick
Only Nemotron 3 Ultra has a coding score, 30.0 on the Vibe Coding Index; Qwen3 235B A22B Instruct 2507 has none, so the two cannot be ranked on coding.
Pick Qwen3 235B A22B Instruct 2507 for output tokens 3.1x cheaper ($0.775 vs $2.40 per million) and work measured by SciCode, where it leads.
- Shared results
- 17
- Nemotron 3 Ultra only
- 54
- Qwen3 235B A22B Instruct 2507 only
- 7
- Like-for-like categories
- 6 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
Only Nemotron 3 Ultra has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Nemotron 3 Ultra
Nemotron 3 Ultra leads, 22.9 against 12.0.
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
No call
Only Nemotron 3 Ultra has a GPQA Diamond result, 86.1%.
Directional only
Knowledge
MMLU Pro
No call
Only Nemotron 3 Ultra has a MMLU Pro result, 85.8%.
Directional only
Web design
Design Arena: websites
Nemotron 3 Ultra
Nemotron 3 Ultra leads, 1145 against 1058.
Same board, both measured
Long documents
Listed context window
No call
Both take 262K tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3 235B A22B Instruct 2507
Nemotron 3 Ultra $0.0018, Qwen3 235B A22B Instruct 2507 $0.0005 per turn.
List prices
At a glance
- Coding index
- Nemotron 3 Ultra 30.0 from 3 benchmarks, Qwen3 235B A22B Instruct 2507 not benchmarked.
- Widest category
- Agentic work, 30.1 points apart on average over 1 shared benchmark.
- Widest single gap
- IFBench: Nemotron 3 Ultra 81.4% against Qwen3 235B A22B Instruct 2507 51.2%, 30.1 points apart on a 0 to 100 scale.
- Output price
- Nemotron 3 Ultra $2.40 against Qwen3 235B A22B Instruct 2507 $0.775 per million output tokens, 3.1x apart.
- Context window
- Both take 262K tokens.
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 | Nemotron 3 Ultra | Bars | Qwen3 235B A22B Instruct 2507 | Ties |
|---|---|---|---|---|
| Coding | 2 | 1 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 2 | 0 | ||
| Web design | 6 | 0 | ||
| General | 4 | 0 | ||
| All shared boards | 16 | 1 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index21
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| AA Coding Index | 49.3 | Not measured | Only Nemotron 3 Ultra measured |
| LiveCodeBench (Vals AI) | 86.0% | Not measured | Only Nemotron 3 Ultra measured |
| SWE-bench Verified (Vals AI) | 69.0% | Not measured | Only Nemotron 3 Ultra measured |
| Vibe Code Bench (Vals AI) | 7.6% | Not measured | Only Nemotron 3 Ultra measured |
| Code migration (Vals AI) | 4.9% | Not measured | Only Nemotron 3 Ultra measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Nemotron 3 Ultra measured |
| Terminal-Bench 2.1 (Vals AI) | 50.9% | Not measured | Only Nemotron 3 Ultra measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 78.8% | Only Qwen3 235B A22B Instruct 2507 measured |
| SciCode | 40.3% | 41.4% | Qwen3 235B A22B Instruct 2507 |
| Terminal-Bench Hard (Artificial Analysis run) | 36.4% | 15.2% | Nemotron 3 Ultra |
| FrontierCode | 13.6% | Not measured | Only Nemotron 3 Ultra measured |
| WeirdML | 43.5% | 38.7% | Nemotron 3 Ultra |
| LMArena Text: coding | 1467 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: software and IT | 1468 | Not measured | Only Nemotron 3 Ultra measured |
| Neither model has a result on the other 29 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| AA Intelligence Index | 22.9 | 12.0 | Nemotron 3 Ultra |
| Epoch Capabilities Index | 146.0 | Not measured | Only Nemotron 3 Ultra measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| AA Agentic Index | 20.1 | Not measured | Only Nemotron 3 Ultra measured |
| Finance Agent (Vals AI) | 37.7% | Not measured | Only Nemotron 3 Ultra measured |
| Legal Research Bench (Vals AI) | 15.4% | Not measured | Only Nemotron 3 Ultra measured |
| Excel Modeling Benchmark (Vals AI) | 31.7% | Not measured | Only Nemotron 3 Ultra measured |
| τ²-Bench Telecom | 83.3% | 53.2% | Nemotron 3 Ultra |
| APEX-Agents | 22.7% | Not measured | Only Nemotron 3 Ultra measured |
| Berkeley Function Calling Leaderboard (BFCL) | Not measured | 52.1% | Only Qwen3 235B A22B Instruct 2507 measured |
| MCP Atlas | 63.1% | Not measured | 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 index20
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| GPQA Diamond | 86.1% | Not measured | Only Nemotron 3 Ultra measured |
| GPQA Diamond (Artificial Analysis run) | 86.7% | 79.0% | Nemotron 3 Ultra |
| Humanity's Last Exam (Artificial Analysis run) | 28.4% | 15.9% | Nemotron 3 Ultra |
| GPQA Diamond (Epoch AI run) | 85.4% | Not measured | Only Nemotron 3 Ultra measured |
| ARC-AGI | Not measured | 11.0% | Only Qwen3 235B A22B Instruct 2507 measured |
| ARC-AGI-2 | Not measured | 1.3% | Only Qwen3 235B A22B Instruct 2507 measured |
| Mystery game puzzles | 20.0% | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: hard prompts | 1452 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: expert | 1472 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: hard prompts in English | 1453 | Not measured | Only Nemotron 3 Ultra measured |
| Neither model has a result on the other 13 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| AA Math Index | Not measured | 91.0 | Only Qwen3 235B A22B Instruct 2507 measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 91.0% | Only Qwen3 235B A22B Instruct 2507 measured |
| OTIS Mock AIME | 86.7% | Not measured | Only Nemotron 3 Ultra measured |
| ProofBench | 2.0% | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: math | 1457 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: mathematical work | 1462 | Not measured | Only Nemotron 3 Ultra measured |
| Neither model has a result on the other 8 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| MMLU Pro | 85.8% | Not measured | Only Nemotron 3 Ultra measured |
| LegalBench (Vals AI) | 82.1% | Not measured | Only Nemotron 3 Ultra measured |
| CorpFin (Vals AI) | 65.5% | Not measured | Only Nemotron 3 Ultra measured |
| TaxEval (Vals AI) | 73.1% | Not measured | Only Nemotron 3 Ultra measured |
| MedCode (Vals AI) | 38.6% | Not measured | Only Nemotron 3 Ultra measured |
| MMLU Pro (Artificial Analysis run) | Not measured | 84.3% | Only Qwen3 235B A22B Instruct 2507 measured |
| LMArena Text: science | 1465 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: medicine and healthcare | 1455 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: legal and government | 1453 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: business and finance | 1435 | Not measured | Only Nemotron 3 Ultra measured |
| Neither model has a result on the other 10 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index60
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| Design Arena: websites | 1145 | 1058 | Nemotron 3 Ultra |
| Design Arena: UI components | 1135 | 963 | Nemotron 3 Ultra |
| Design Arena: game dev | 1138 | 951 | Nemotron 3 Ultra |
| Design Arena: data viz | 1142 | 1070 | Nemotron 3 Ultra |
| Design Arena: 3D | 1132 | 995 | Nemotron 3 Ultra |
| Design Arena: all code categories | 1146 | 1038 | Nemotron 3 Ultra |
| Design Arena: SVG | 1069 | Not measured | Only Nemotron 3 Ultra measured |
| Design Arena: ASCII art | 1100 | Not measured | Only Nemotron 3 Ultra measured |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| IFBench | 81.4% | 51.2% | Nemotron 3 Ultra |
| Long context reasoning (AA-LCR) | 79.3% | 72.0% | Nemotron 3 Ultra |
| DTBench | 90.1% | 78.4% | Nemotron 3 Ultra |
| LMCA | 36.9% | 23.5% | Nemotron 3 Ultra |
| LMArena Text | 1445 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: instruction following | 1418 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: creative writing | 1398 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: English | 1454 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: non-English | 1427 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: Chinese | 1498 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: German | 1474 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: Korean | 1385 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: French | 1473 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: Spanish | 1454 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: Russian | 1417 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: Polish | 1385 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: multi-turn | 1411 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: longer queries | 1437 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: writing and language | 1405 | Not measured | Only Nemotron 3 Ultra measured |
| LMArena Text: entertainment and media | 1397 | Not measured | Only Nemotron 3 Ultra measured |
| Neither model has a result on the other 28 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 | 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
Qwen3 235B A22B Instruct 2507 costs less
Repository review
50K input + 3K output tokens
Qwen3 235B A22B Instruct 2507 costs less
Agent loop
200K input + 10K output tokens
Qwen3 235B A22B Instruct 2507 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 | NVIDIA Nemotron 3 Ultra | Qwen3 235B A22B Instruct 2507 |
|---|---|---|
| Maker | NVIDIA | Qwen |
| Weights | Open weights | Open weights |
| Released | June 4, 2026 | July 21, 2025 |
| Context window | 262K | 262K |
| Input $/Mtok | $0.60 | $0.145 |
| Output $/Mtok | $2.40 | $0.775 |
| Coding index | 30.0 | Not benchmarked |
| Independent measurements behind it | 3 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Nemotron 3 Ultra better than Qwen3 235B A22B Instruct 2507 for coding?
Only Nemotron 3 Ultra has a coding score, 30.0 on the Vibe Coding Index; Qwen3 235B A22B Instruct 2507 has none, so the two cannot be ranked on coding. Pick Qwen3 235B A22B Instruct 2507 for output tokens 3.1x cheaper ($0.775 vs $2.40 per million) and work measured by SciCode, where it leads.
Which is cheaper, Nemotron 3 Ultra or Qwen3 235B A22B Instruct 2507?
Nemotron 3 Ultra costs $0.60 per million input tokens and $2.40 per million output tokens; Qwen3 235B A22B Instruct 2507 costs $0.145 and $0.775. Cost per solved task is known for neither of them.
Which has the bigger context window, Nemotron 3 Ultra or Qwen3 235B A22B Instruct 2507?
Neither: both take 262K tokens.
How many benchmarks is the Nemotron 3 Ultra vs Qwen3 235B A22B Instruct 2507 comparison based on?
17 benchmarks have a result for both models as of October 23, 2026. Nemotron 3 Ultra's coding index rests on 3 independent benchmarks and Qwen3 235B A22B Instruct 2507's on 0. Where only one model has a result, the page says so instead of naming a winner.