DeepSeek V4 Flash 0731 vs Muse Spark 1.3
Muse Spark 1.3 leads DeepSeek V4 Flash 0731 on coding, 57.4 to 51.8 on the Vibe Coding Index, and DeepSeek V4 Flash 0731 costs 10.7x less per output token ($0.396 vs $4.25 per million). Updated October 22, 2026; 37 benchmarks have a result for both.
DeepSeek V4 Flash 0731 or Muse Spark 1.3: which to pick
Pick Muse Spark 1.3 for coding: it scores 57.4 on the Vibe Coding Index against 51.8 for DeepSeek V4 Flash 0731.
Pick DeepSeek V4 Flash 0731 for output tokens 10.7x cheaper ($0.396 vs $4.25 per million) and work measured by Terminal-Bench 4 (Vals AI) or Code migration (Vals AI), where it leads.
- Shared results
- 37
- DeepSeek V4 Flash 0731 only
- 18
- Muse Spark 1.3 only
- 55
- Like-for-like categories
- 7 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
Muse Spark 1.3
Muse Spark 1.3 scores 57.4 against 51.8.
4 benchmarks behind each score
Intelligence
AA Intelligence Index
Muse Spark 1.3
Muse Spark 1.3 leads, 48.1 against 34.3.
Same board, both measured
Agentic work
AA Agentic Index
Muse Spark 1.3
Muse Spark 1.3 leads, 55.5 against 41.0.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only DeepSeek V4 Flash 0731 has a GPQA Diamond result, 89.9%.
Directional only
Knowledge
MMLU Pro
No call
Only DeepSeek V4 Flash 0731 has a MMLU Pro result, 86.2%.
Directional only
Web design
Design Arena: websites
Muse Spark 1.3
Muse Spark 1.3 leads, 1356 against 1237.
Same board, both measured
Long documents
Listed context window
No call
Both take 1.05M tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
DeepSeek V4 Flash 0731
DeepSeek V4 Flash 0731 $0.0003, Muse Spark 1.3 $0.0034 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only DeepSeek V4 Flash 0731 has the number, $0.01.
Directional only
At a glance
- Coding index
- DeepSeek V4 Flash 0731 51.8 from 4 benchmarks, Muse Spark 1.3 57.4 from 4 benchmarks.
- Widest category
- Web design, 17.1 points apart on average over 8 shared benchmarks.
- Widest single gap
- Design Arena: ASCII art: DeepSeek V4 Flash 0731 1090 against Muse Spark 1.3 1278, 23.5 points apart on a 0 to 100 scale.
- Output price
- DeepSeek V4 Flash 0731 $0.396 against Muse Spark 1.3 $4.25 per million output tokens, 10.7x apart.
- Context window
- Both take 1.05M 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 | DeepSeek V4 Flash 0731 | Bars | Muse Spark 1.3 | Ties |
|---|---|---|---|---|
| Coding | 3 | 7 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 5 | 1 tie | |
| Reasoning | 1 | 3 | ||
| Math | 0 | 4 | ||
| Web design | 0 | 8 | ||
| General | 0 | 3 | ||
| All shared boards | 4 | 32 | 1 tie |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index37
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| LiveBench coding | Not measured | 90.2% | Only Muse Spark 1.3 measured |
| LMArena WebDev | Not measured | 1659 | Only Muse Spark 1.3 measured |
| AA Coding Index | 69.1 | 75.8 | Muse Spark 1.3 |
| LiveCodeBench (Vals AI) | 87.3% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| SWE-bench Verified (Vals AI) | 88.8% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| Terminal-Bench 4 (Vals AI) | 18.7% | 10.6% | DeepSeek V4 Flash 0731 |
| Vibe Code Bench (Vals AI) | 74.7% | 82.9% | Muse Spark 1.3 |
| Code migration (Vals AI) | 38.6% | 27.6% | DeepSeek V4 Flash 0731 |
| ProgramBench (Vals AI) | 0.0% | 0.5% | Muse Spark 1.3 |
| IOI (Vals AI) | 32.7% | 43.9% | Muse Spark 1.3 |
| Terminal-Bench 2.1 (Vals AI) | 67.0% | 72.3% | Muse Spark 1.3 |
| SkillsBench (Vals AI) | 50.7% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| CyberBench (Vals AI) | 72.1% | 69.4% | DeepSeek V4 Flash 0731 |
| SciCode | 50.3% | 59.7% | Muse Spark 1.3 |
| FrontierCode | 18.8% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| FrontierSWE | Not measured | 25.8% | Only Muse Spark 1.3 measured |
| WeirdML | 63.0% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| ALE-Bench | 1306.1 | 1688.3 | Muse Spark 1.3 |
| LMArena Text: coding | Not measured | 1510 | Only Muse Spark 1.3 measured |
| LMArena Text: software and IT | Not measured | 1511 | Only Muse Spark 1.3 measured |
| LMArena WebDev: React | Not measured | 1672 | Only Muse Spark 1.3 measured |
| LMArena WebDev: HTML | Not measured | 1653 | Only Muse Spark 1.3 measured |
| LMArena WebDev: from an image | Not measured | 1640 | Only Muse Spark 1.3 measured |
| Neither model has a result on the other 16 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| AA Intelligence Index | 34.3 | 48.1 | Muse Spark 1.3 |
| Epoch Capabilities Index | 154.8 | 157.2 | Muse Spark 1.3 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index05
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| AA Agentic Index | 41.0 | 55.5 | Muse Spark 1.3 |
| Design Arena: agents building web apps | Not measured | 1297 | Only Muse Spark 1.3 measured |
| Design Arena: agents building full-stack apps | Not measured | 1295 | Only Muse Spark 1.3 measured |
| Design Arena: agents building mobile apps | Not measured | 1218 | Only Muse Spark 1.3 measured |
| Design Arena: agents building Godot games | Not measured | 1241 | Only Muse Spark 1.3 measured |
| Design Arena: PowerPoint slides | Not measured | 1219 | Only Muse Spark 1.3 measured |
| Design Arena: spreadsheets | Not measured | 1062 | Only Muse Spark 1.3 measured |
| LiveBench research | Not measured | 68.6% | Only Muse Spark 1.3 measured |
| LiveBench browser | Not measured | 80.0% | Only Muse Spark 1.3 measured |
| LiveBench long agentic | Not measured | 40.4% | Only Muse Spark 1.3 measured |
| Finance Agent (Vals AI) | 49.5% | 58.9% | Muse Spark 1.3 |
| Tax Agent Bench (Vals AI) | 28.1% | 36.0% | Muse Spark 1.3 |
| Legal Research Bench (Vals AI) | 30.3% | 40.9% | Muse Spark 1.3 |
| Excel Modeling Benchmark (Vals AI) | 57.0% | 62.7% | Muse Spark 1.3 |
| BioMysteryBench (Vals AI) | 64.4% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| Terminal-Bench-Science (Vals AI) | 4.3% | 4.3% | Tie |
| APEX-Agents | Not measured | 58.6% | Only Muse Spark 1.3 measured |
| Neither model has a result on the other 24 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index13
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| GPQA Diamond | 89.9% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| LiveBench reasoning | Not measured | 89.7% | Only Muse Spark 1.3 measured |
| GPQA Diamond (Artificial Analysis run) | 90.8% | 94.1% | Muse Spark 1.3 |
| Humanity's Last Exam (Artificial Analysis run) | 38.6% | 48.7% | Muse Spark 1.3 |
| GPQA Diamond (Epoch AI run) | 91.0% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| ARC-AGI | 89.0% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| ARC-AGI-2 | 61.4% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| SimpleBench | 61.1% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| Chess puzzles | 32.0% | 37.0% | Muse Spark 1.3 |
| Mystery game puzzles | 34.0% | 25.0% | DeepSeek V4 Flash 0731 |
| LMArena Text: hard prompts | Not measured | 1503 | Only Muse Spark 1.3 measured |
| LMArena Text: expert | Not measured | 1519 | Only Muse Spark 1.3 measured |
| LMArena Text: hard prompts in English | Not measured | 1500 | Only Muse Spark 1.3 measured |
| LMArena Vision: diagrams | Not measured | 1330 | Only Muse Spark 1.3 measured |
| LMArena Vision: homework | Not measured | 1296 | Only Muse Spark 1.3 measured |
| Neither model has a result on the other 7 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index04
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| OTIS Mock AIME | 94.4% | 99.2% | Muse Spark 1.3 |
| FrontierMath, tiers 1 to 3 | 57.5% | 74.4% | Muse Spark 1.3 |
| FrontierMath, tier 4 | 24.4% | 46.3% | Muse Spark 1.3 |
| ProofBench | 56.0% | 58.0% | Muse Spark 1.3 |
| LMArena Text: math | Not measured | 1497 | Only Muse Spark 1.3 measured |
| LMArena Text: mathematical work | Not measured | 1491 | Only Muse Spark 1.3 measured |
| Neither model has a result on the other 7 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| MMLU Pro | 86.2% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| LegalBench (Vals AI) | 77.7% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| CorpFin (Vals AI) | 61.9% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| TaxEval (Vals AI) | 70.7% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| MedCode (Vals AI) | 41.4% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| SimpleQA Verified | 33.6% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| LMArena Text: science | Not measured | 1507 | Only Muse Spark 1.3 measured |
| LMArena Text: medicine and healthcare | Not measured | 1493 | Only Muse Spark 1.3 measured |
| LMArena Text: legal and government | Not measured | 1499 | Only Muse Spark 1.3 measured |
| LMArena Text: business and finance | Not measured | 1490 | Only Muse Spark 1.3 measured |
| Neither model has a result on the other 8 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index08
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| Design Arena: websites | 1237 | 1356 | Muse Spark 1.3 |
| Design Arena: UI components | 1233 | 1349 | Muse Spark 1.3 |
| Design Arena: game dev | 1207 | 1328 | Muse Spark 1.3 |
| Design Arena: data viz | 1180 | 1343 | Muse Spark 1.3 |
| Design Arena: 3D | 1208 | 1358 | Muse Spark 1.3 |
| Design Arena: all code categories | 1229 | 1352 | Muse Spark 1.3 |
| Design Arena: SVG | 1184 | 1301 | Muse Spark 1.3 |
| Design Arena: ASCII art | 1090 | 1278 | Muse Spark 1.3 |
| LiveBench design | Not measured | 74.8% | Only Muse Spark 1.3 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index03
| Benchmark | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | Leader |
|---|---|---|---|
| LiveBench data analysis | Not measured | 79.6% | Only Muse Spark 1.3 measured |
| MedScribe (Vals AI) | 80.4% | Not measured | Only DeepSeek V4 Flash 0731 measured |
| Long context reasoning (AA-LCR) | 79.7% | 83.0% | Muse Spark 1.3 |
| DTBench | 90.9% | 96.5% | Muse Spark 1.3 |
| LMCA | 41.7% | 53.9% | Muse Spark 1.3 |
| LMArena Text | Not measured | 1490 | Only Muse Spark 1.3 measured |
| LMArena Text: instruction following | Not measured | 1479 | Only Muse Spark 1.3 measured |
| LMArena Text: creative writing | Not measured | 1462 | Only Muse Spark 1.3 measured |
| LMArena Vision | Not measured | 1308 | Only Muse Spark 1.3 measured |
| LMArena Text: English | Not measured | 1495 | Only Muse Spark 1.3 measured |
| LMArena Text: non-English | Not measured | 1481 | Only Muse Spark 1.3 measured |
| LMArena Text: Chinese | Not measured | 1532 | Only Muse Spark 1.3 measured |
| LMArena Text: German | Not measured | 1513 | Only Muse Spark 1.3 measured |
| LMArena Text: Japanese | Not measured | 1470 | Only Muse Spark 1.3 measured |
| LMArena Text: Korean | Not measured | 1490 | Only Muse Spark 1.3 measured |
| LMArena Text: French | Not measured | 1511 | Only Muse Spark 1.3 measured |
| LMArena Text: Spanish | Not measured | 1496 | Only Muse Spark 1.3 measured |
| LMArena Text: Russian | Not measured | 1491 | Only Muse Spark 1.3 measured |
| LMArena Text: Polish | Not measured | 1490 | Only Muse Spark 1.3 measured |
| LMArena Text: multi-turn | Not measured | 1487 | Only Muse Spark 1.3 measured |
| LMArena Text: longer queries | Not measured | 1488 | Only Muse Spark 1.3 measured |
| LMArena Text: writing and language | Not measured | 1470 | Only Muse Spark 1.3 measured |
| LMArena Text: entertainment and media | Not measured | 1457 | Only Muse Spark 1.3 measured |
| LMArena Vision: English | Not measured | 1298 | Only Muse Spark 1.3 measured |
| LMArena Vision: Chinese | Not measured | 1384 | Only Muse Spark 1.3 measured |
| LMArena Vision: reading text | Not measured | 1319 | Only Muse Spark 1.3 measured |
| LMArena Vision: humor | Not measured | 1293 | Only Muse Spark 1.3 measured |
| LMArena Vision: creative writing | Not measured | 1296 | Only Muse Spark 1.3 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 | DeepSeek V4 Flash 0731 | Muse Spark 1.3 | 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 V4 Flash 0731 costs less
Repository review
50K input + 3K output tokens
DeepSeek V4 Flash 0731 costs less
Agent loop
200K input + 10K output tokens
DeepSeek V4 Flash 0731 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 V4 Flash 0731 | Meta Muse Spark 1.3 |
|---|---|---|
| Maker | DeepSeek | Meta |
| Weights | Open weights | No public weights listed |
| Released | July 31, 2026 | September 2, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $0.14 | $1.25 |
| Output $/Mtok | $0.396 | $4.25 |
| Coding index | 51.8 | 57.4 |
| Independent measurements behind it | 4 | 4 |
| Cost per solved task | $0.01 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is DeepSeek V4 Flash 0731 better than Muse Spark 1.3 for coding?
Pick Muse Spark 1.3 for coding: it scores 57.4 on the Vibe Coding Index against 51.8 for DeepSeek V4 Flash 0731. Pick DeepSeek V4 Flash 0731 for output tokens 10.7x cheaper ($0.396 vs $4.25 per million) and work measured by Terminal-Bench 4 (Vals AI) or Code migration (Vals AI), where it leads.
Which is cheaper, DeepSeek V4 Flash 0731 or Muse Spark 1.3?
DeepSeek V4 Flash 0731 costs $0.14 per million input tokens and $0.396 per million output tokens; Muse Spark 1.3 costs $1.25 and $4.25. Cost per solved task is known for DeepSeek V4 Flash 0731 only.
Which has the bigger context window, DeepSeek V4 Flash 0731 or Muse Spark 1.3?
Neither: both take 1.05M tokens.
How many benchmarks is the DeepSeek V4 Flash 0731 vs Muse Spark 1.3 comparison based on?
37 benchmarks have a result for both models as of October 22, 2026. DeepSeek V4 Flash 0731's coding index rests on 4 independent benchmarks and Muse Spark 1.3's on 4. Where only one model has a result, the page says so instead of naming a winner.