Muse Spark 1.2 vs Qwen3 235B A22B Thinking 2507
Muse Spark 1.2 leads Qwen3 235B A22B Thinking 2507 on the Artificial Analysis Intelligence Index, 39.6 to 12.7, and Qwen3 235B A22B Thinking 2507 costs 1.2x less per output token ($3.50 vs $4.25 per million). Updated October 22, 2026; 37 benchmarks have a result for both.
Muse Spark 1.2 or Qwen3 235B A22B Thinking 2507: which to pick
Only Muse Spark 1.2 has a coding score, 55.7 on the Vibe Coding Index; Qwen3 235B A22B Thinking 2507 has none, so the two cannot be ranked on coding.
Pick Qwen3 235B A22B Thinking 2507 for output tokens 1.2x cheaper ($3.50 vs $4.25 per million).
- Shared results
- 37
- Muse Spark 1.2 only
- 60
- Qwen3 235B A22B Thinking 2507 only
- 9
- 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
Only Muse Spark 1.2 has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Muse Spark 1.2
Muse Spark 1.2 leads, 39.6 against 12.7.
Same board, both measured
Agentic work
AA Agentic Index
Muse Spark 1.2
Muse Spark 1.2 leads, 43.2 against 1.3.
Same board, both measured
Knowledge
MMLU Pro
No call
Only Muse Spark 1.2 has a MMLU Pro result, 88.3%.
Directional only
Web design
Design Arena: websites
Muse Spark 1.2
Muse Spark 1.2 leads, 1312 against 1052.
Same board, both measured
Long documents
Listed context window
Muse Spark 1.2
Muse Spark 1.2 takes 1.05M tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3 235B A22B Thinking 2507
Muse Spark 1.2 $0.0034, Qwen3 235B A22B Thinking 2507 $0.0022 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Muse Spark 1.2 has the number, $0.64.
Directional only
At a glance
- Coding index
- Muse Spark 1.2 55.7 from 5 benchmarks, Qwen3 235B A22B Thinking 2507 not benchmarked.
- Widest category
- Agentic work, 41.9 points apart on average over 1 shared benchmark.
- Widest single gap
- AA Coding Index: Muse Spark 1.2 72.2 against Qwen3 235B A22B Thinking 2507 22.1, 50.1 points apart on a 0 to 100 scale.
- Output price
- Muse Spark 1.2 $4.25 against Qwen3 235B A22B Thinking 2507 $3.50 per million output tokens, 1.2x apart.
- Context window
- Muse Spark 1.2 takes 1.05M tokens, Qwen3 235B A22B Thinking 2507 128K.
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 | Muse Spark 1.2 | Bars | Qwen3 235B A22B Thinking 2507 | Ties |
|---|---|---|---|---|
| Coding | 4 | 0 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 3 | 0 | ||
| Math | 2 | 0 | ||
| Knowledge | 5 | 0 | ||
| Web design | 6 | 0 | ||
| General | 15 | 0 | ||
| All shared boards | 37 | 0 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index40
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| LiveBench coding | 78.2% | Not measured | Only Muse Spark 1.2 measured |
| LMArena WebDev | 1534 | Not measured | Only Muse Spark 1.2 measured |
| AA Coding Index | 72.2 | 22.1 | Muse Spark 1.2 |
| SWE-bench Verified (Vals AI) | 86.6% | Not measured | Only Muse Spark 1.2 measured |
| Terminal-Bench 4 (Vals AI) | 6.1% | Not measured | Only Muse Spark 1.2 measured |
| Vibe Code Bench (Vals AI) | 79.1% | Not measured | Only Muse Spark 1.2 measured |
| Code migration (Vals AI) | 29.9% | Not measured | Only Muse Spark 1.2 measured |
| ProgramBench (Vals AI) | 0.5% | Not measured | Only Muse Spark 1.2 measured |
| IOI (Vals AI) | 21.8% | Not measured | Only Muse Spark 1.2 measured |
| Terminal-Bench 2.1 (Vals AI) | 69.7% | Not measured | Only Muse Spark 1.2 measured |
| SkillsBench (Vals AI) | 53.0% | Not measured | Only Muse Spark 1.2 measured |
| CyberBench (Vals AI) | 69.5% | Not measured | Only Muse Spark 1.2 measured |
| SciCode | 57.4% | Not measured | Only Muse Spark 1.2 measured |
| FrontierSWE | 12.0% | Not measured | Only Muse Spark 1.2 measured |
| WeirdML | 60.3% | 41.0% | Muse Spark 1.2 |
| LMArena Text: coding | 1505 | 1424 | Muse Spark 1.2 |
| LMArena Text: software and IT | 1505 | 1431 | Muse Spark 1.2 |
| LMArena WebDev: React | 1526 | Not measured | Only Muse Spark 1.2 measured |
| LMArena WebDev: HTML | 1543 | Not measured | Only Muse Spark 1.2 measured |
| Neither model has a result on the other 24 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| AA Intelligence Index | 39.6 | 12.7 | Muse Spark 1.2 |
| Epoch Capabilities Index | 155.1 | Not measured | Only Muse Spark 1.2 measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| AA Agentic Index | 43.2 | 1.3 | Muse Spark 1.2 |
| Design Arena: agents building web apps | 1221 | Not measured | Only Muse Spark 1.2 measured |
| Design Arena: agents building full-stack apps | 1210 | Not measured | Only Muse Spark 1.2 measured |
| Design Arena: agents building mobile apps | 1181 | Not measured | Only Muse Spark 1.2 measured |
| Design Arena: agents building Android apps | 1222 | Not measured | Only Muse Spark 1.2 measured |
| Design Arena: agents building games | 1178 | Not measured | Only Muse Spark 1.2 measured |
| Design Arena: HTML slides | 1074 | Not measured | Only Muse Spark 1.2 measured |
| Design Arena: PowerPoint slides | 1170 | Not measured | Only Muse Spark 1.2 measured |
| LiveBench research | 67.7% | Not measured | Only Muse Spark 1.2 measured |
| LiveBench browser | 36.0% | Not measured | Only Muse Spark 1.2 measured |
| LiveBench long agentic | 21.3% | Not measured | Only Muse Spark 1.2 measured |
| SREBench (Vals AI) | 0.0% | Not measured | Only Muse Spark 1.2 measured |
| Finance Agent (Vals AI) | 60.6% | Not measured | Only Muse Spark 1.2 measured |
| Tax Agent Bench (Vals AI) | 32.8% | Not measured | Only Muse Spark 1.2 measured |
| Legal Research Bench (Vals AI) | 43.8% | Not measured | Only Muse Spark 1.2 measured |
| Excel Modeling Benchmark (Vals AI) | 57.0% | Not measured | Only Muse Spark 1.2 measured |
| BioMysteryBench (Vals AI) | 64.8% | Not measured | Only Muse Spark 1.2 measured |
| APEX-Agents | 45.4% | Not measured | Only Muse Spark 1.2 measured |
| Vending-Bench 2 | Not measured | -$11 | Only Qwen3 235B A22B Thinking 2507 measured |
| Neither model has a result on the other 23 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| LiveBench reasoning | 90.0% | Not measured | Only Muse Spark 1.2 measured |
| MMMU Pro (Vals AI) | 86.1% | Not measured | Only Muse Spark 1.2 measured |
| GPQA Diamond (Artificial Analysis run) | 90.4% | Not measured | Only Muse Spark 1.2 measured |
| Humanity's Last Exam (Artificial Analysis run) | 45.5% | Not measured | Only Muse Spark 1.2 measured |
| GPQA Diamond (Epoch AI run) | Not measured | 80.1% | Only Qwen3 235B A22B Thinking 2507 measured |
| SimpleBench | 74.5% | Not measured | Only Muse Spark 1.2 measured |
| Chess puzzles | 27.0% | Not measured | Only Muse Spark 1.2 measured |
| Mystery game puzzles | Not measured | 9.0% | Only Qwen3 235B A22B Thinking 2507 measured |
| LMArena Text: hard prompts | 1489 | 1416 | Muse Spark 1.2 |
| LMArena Text: expert | 1475 | 1463 | Muse Spark 1.2 |
| LMArena Text: hard prompts in English | 1487 | 1432 | Muse Spark 1.2 |
| LMArena Vision: diagrams | 1317 | Not measured | Only Muse Spark 1.2 measured |
| Neither model has a result on the other 11 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| OTIS Mock AIME | Not measured | 86.7% | Only Qwen3 235B A22B Thinking 2507 measured |
| ProofBench | 43.0% | Not measured | Only Muse Spark 1.2 measured |
| LMArena Text: math | 1484 | 1412 | Muse Spark 1.2 |
| LMArena Text: mathematical work | 1480 | 1435 | Muse Spark 1.2 |
| Neither model has a result on the other 10 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index50
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| MMLU Pro | 88.3% | Not measured | Only Muse Spark 1.2 measured |
| LegalBench (Vals AI) | 85.3% | Not measured | Only Muse Spark 1.2 measured |
| CorpFin (Vals AI) | 70.9% | Not measured | Only Muse Spark 1.2 measured |
| TaxEval (Vals AI) | 80.4% | Not measured | Only Muse Spark 1.2 measured |
| MortgageTax (Vals AI) | 65.4% | Not measured | Only Muse Spark 1.2 measured |
| MedCode (Vals AI) | 49.4% | Not measured | Only Muse Spark 1.2 measured |
| SimpleQA Verified | 60.3% | 40.4% | Muse Spark 1.2 |
| LMArena Text: science | 1500 | 1433 | Muse Spark 1.2 |
| LMArena Text: medicine and healthcare | 1496 | 1456 | Muse Spark 1.2 |
| LMArena Text: legal and government | 1517 | 1415 | Muse Spark 1.2 |
| LMArena Text: business and finance | 1484 | 1413 | Muse Spark 1.2 |
| Neither model has a result on the other 9 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index60
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| Design Arena: websites | 1312 | 1052 | Muse Spark 1.2 |
| Design Arena: UI components | 1299 | 942 | Muse Spark 1.2 |
| Design Arena: game dev | 1300 | 956 | Muse Spark 1.2 |
| Design Arena: data viz | 1334 | 950 | Muse Spark 1.2 |
| Design Arena: 3D | 1300 | 997 | Muse Spark 1.2 |
| Design Arena: all code categories | 1312 | 1032 | Muse Spark 1.2 |
| Design Arena: SVG | 1283 | Not measured | Only Muse Spark 1.2 measured |
| Design Arena: ASCII art | 1284 | Not measured | Only Muse Spark 1.2 measured |
| LiveBench design | 67.1% | Not measured | Only Muse Spark 1.2 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index150
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| LiveBench data analysis | 76.5% | Not measured | Only Muse Spark 1.2 measured |
| Public Benefits Bench (Vals AI) | 68.5% | Not measured | Only Muse Spark 1.2 measured |
| SAGE (Vals AI) | 47.7% | Not measured | Only Muse Spark 1.2 measured |
| MedScribe (Vals AI) | 90.1% | Not measured | Only Muse Spark 1.2 measured |
| Vals Multimodal Index | 69.8% | Not measured | Only Muse Spark 1.2 measured |
| Long context reasoning (AA-LCR) | 79.0% | Not measured | Only Muse Spark 1.2 measured |
| DTBench | 94.7% | 80.3% | Muse Spark 1.2 |
| LMCA | 48.4% | 29.3% | Muse Spark 1.2 |
| Fiction.LiveBench | Not measured | 75.0% | Only Qwen3 235B A22B Thinking 2507 measured |
| Creative writing (Lech Mazur) | Not measured | 82.4% | Only Qwen3 235B A22B Thinking 2507 measured |
| LMArena Text | 1482 | 1415 | Muse Spark 1.2 |
| LMArena Text: instruction following | 1462 | 1386 | Muse Spark 1.2 |
| LMArena Text: creative writing | 1436 | 1384 | Muse Spark 1.2 |
| LMArena Vision | 1304 | Not measured | Only Muse Spark 1.2 measured |
| LMArena Text: English | 1483 | 1429 | Muse Spark 1.2 |
| LMArena Text: non-English | 1476 | 1397 | Muse Spark 1.2 |
| LMArena Text: Chinese | 1509 | 1481 | Muse Spark 1.2 |
| LMArena Text: German | Not measured | 1408 | Only Qwen3 235B A22B Thinking 2507 measured |
| LMArena Text: Japanese | Not measured | 1386 | Only Qwen3 235B A22B Thinking 2507 measured |
| LMArena Text: Korean | 1462 | 1375 | Muse Spark 1.2 |
| LMArena Text: French | 1495 | Not measured | Only Muse Spark 1.2 measured |
| LMArena Text: Spanish | 1484 | 1389 | Muse Spark 1.2 |
| LMArena Text: Russian | 1483 | 1401 | Muse Spark 1.2 |
| LMArena Text: Polish | Not measured | 1390 | Only Qwen3 235B A22B Thinking 2507 measured |
| LMArena Text: multi-turn | 1502 | 1406 | Muse Spark 1.2 |
| LMArena Text: longer queries | 1473 | 1400 | Muse Spark 1.2 |
| LMArena Text: writing and language | 1453 | 1388 | Muse Spark 1.2 |
| LMArena Text: entertainment and media | 1443 | 1370 | Muse Spark 1.2 |
| LMArena Vision: English | 1291 | Not measured | Only Muse Spark 1.2 measured |
| LMArena Vision: reading text | 1313 | Not measured | Only Muse Spark 1.2 measured |
| Neither model has a result on the other 18 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Muse Spark 1.2 | Qwen3 235B A22B Thinking 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 Thinking 2507 costs less
Repository review
50K input + 3K output tokens
Qwen3 235B A22B Thinking 2507 costs less
Agent loop
200K input + 10K output tokens
Qwen3 235B A22B Thinking 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 | Meta Muse Spark 1.2 | Qwen3 235B A22B Thinking 2507 |
|---|---|---|
| Maker | Meta | Qwen |
| Weights | No public weights listed | Open weights |
| Released | August 5, 2026 | July 25, 2025 |
| Context window | 1.05M | 128K |
| Input $/Mtok | $1.25 | $0.45 |
| Output $/Mtok | $4.25 | $3.50 |
| Coding index | 55.7 | Not benchmarked |
| Independent measurements behind it | 5 | 0 |
| Cost per solved task | $0.64 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Muse Spark 1.2 better than Qwen3 235B A22B Thinking 2507 for coding?
Only Muse Spark 1.2 has a coding score, 55.7 on the Vibe Coding Index; Qwen3 235B A22B Thinking 2507 has none, so the two cannot be ranked on coding. Pick Qwen3 235B A22B Thinking 2507 for output tokens 1.2x cheaper ($3.50 vs $4.25 per million).
Which is cheaper, Muse Spark 1.2 or Qwen3 235B A22B Thinking 2507?
Muse Spark 1.2 costs $1.25 per million input tokens and $4.25 per million output tokens; Qwen3 235B A22B Thinking 2507 costs $0.45 and $3.50. Cost per solved task is known for Muse Spark 1.2 only.
Which has the bigger context window, Muse Spark 1.2 or Qwen3 235B A22B Thinking 2507?
Muse Spark 1.2, with 1.05M tokens against 128K.
How many benchmarks is the Muse Spark 1.2 vs Qwen3 235B A22B Thinking 2507 comparison based on?
37 benchmarks have a result for both models as of October 22, 2026. Muse Spark 1.2's coding index rests on 5 independent benchmarks and Qwen3 235B A22B Thinking 2507's on 0. Where only one model has a result, the page says so instead of naming a winner.