DeepSeek V3.2 vs DeepSeek V4.1 Flash
DeepSeek V4.1 Flash leads DeepSeek V3.2 on coding, 62.5 to 30.0 on the Vibe Coding Index, at about the same output price ($0.84 vs $0.88 per million). Updated October 22, 2026; 50 benchmarks have a result for both.
vs boards won of 50 shared
DeepSeek V3.2 or DeepSeek V4.1 Flash: which to pick
Pick DeepSeek V4.1 Flash for coding: it scores 62.5 on the Vibe Coding Index against 30.0 for DeepSeek V3.2.
DeepSeek V3.2 is neither cheaper nor larger in context, and leads no benchmark both models were measured on.
Model ADeepSeek V3.2DeepSeek30.0/1003 independent benchmarks$0.99 per solved taskModel BDeepSeek V4.1 FlashDeepSeek62.5/1005 independent benchmarksNo cost per solved task
- Shared results
- 50
- DeepSeek V3.2 only
- 25
- DeepSeek V4.1 Flash only
- 51
- 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
DeepSeek V4.1 Flash
DeepSeek V4.1 Flash scores 62.5 against 30.0.
3 and 5 benchmarks behind the scores
Intelligence
AA Intelligence Index
DeepSeek V4.1 Flash
DeepSeek V4.1 Flash leads, 24.7 against 21.5.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only DeepSeek V3.2 has a GPQA Diamond result, 80.3%.
Directional only
Math
AIME
No call
Only DeepSeek V3.2 has a AIME result, 84.6%.
Directional only
Knowledge
MMLU Pro
No call
Only DeepSeek V3.2 has a MMLU Pro result, 84.9%.
Directional only
Web design
Design Arena: websites
DeepSeek V4.1 Flash
DeepSeek V4.1 Flash leads, 1308 against 1175.
Same board, both measured
Long documents
Listed context window
DeepSeek V4.1 Flash
DeepSeek V4.1 Flash takes 1.05M tokens against 164K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
DeepSeek V4.1 Flash
DeepSeek V3.2 $0.0007, DeepSeek V4.1 Flash $0.0006 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, DeepSeek V4.1 Flash 62.5 from 5 benchmarks.
- Widest category
- Coding, 35.3 points apart on average over 6 shared benchmarks.
- Widest single gap
- Vibe Code Bench (Vals AI): DeepSeek V3.2 5.1% against DeepSeek V4.1 Flash 84.7%, 79.6 points apart on a 0 to 100 scale.
- Output price
- DeepSeek V3.2 $0.88 against DeepSeek V4.1 Flash $0.84 per million output tokens, 1.0x apart.
- Context window
- DeepSeek V3.2 takes 164K tokens, DeepSeek V4.1 Flash 1.05M.
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 | DeepSeek V4.1 Flash | Ties |
|---|---|---|---|---|
| Coding | 0 | 6 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 1 | ||
| Reasoning | 0 | 6 | ||
| Math | 0 | 2 | ||
| Knowledge | 0 | 5 | ||
| Web design | 0 | 8 | ||
| General | 0 | 20 | ||
| All shared boards | 0 | 50 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index06
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | 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 |
| LiveBench coding | Not measured | 90.2% | Only DeepSeek V4.1 Flash measured |
| LMArena WebDev | 1362 | 1618 | DeepSeek V4.1 Flash |
| AA Coding Index | 44.2 | Not measured | Only DeepSeek V3.2 measured |
| LiveCodeBench (Vals AI) | 80.7% | Not measured | Only DeepSeek V3.2 measured |
| SWE-bench Verified (Vals AI) | 67.6% | Not measured | Only DeepSeek V3.2 measured |
| Terminal-Bench 4 (Vals AI) | Not measured | 19.7% | Only DeepSeek V4.1 Flash measured |
| Vibe Code Bench (Vals AI) | 5.1% | 84.7% | DeepSeek V4.1 Flash |
| Code migration (Vals AI) | Not measured | 45.6% | Only DeepSeek V4.1 Flash measured |
| ProgramBench (Vals AI) | Not measured | 0.5% | Only DeepSeek V4.1 Flash measured |
| IOI (Vals AI) | Not measured | 40.3% | Only DeepSeek V4.1 Flash measured |
| Terminal-Bench 2.1 (Vals AI) | Not measured | 74.5% | Only DeepSeek V4.1 Flash measured |
| SWE-bench Multimodal (Vals AI) | Not measured | 46.5% | Only DeepSeek V4.1 Flash measured |
| Vibe Code Bench 1-100 (Vals AI) | Not measured | 16.4% | Only DeepSeek V4.1 Flash measured |
| SkillsBench (Vals AI) | Not measured | 69.8% | Only DeepSeek V4.1 Flash measured |
| CyberBench (Vals AI) | Not measured | 73.7% | Only DeepSeek V4.1 Flash measured |
| LiveCodeBench (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| SciCode | Not measured | 51.9% | Only DeepSeek V4.1 Flash measured |
| Terminal-Bench Hard (Artificial Analysis run) | 35.6% | Not measured | Only DeepSeek V3.2 measured |
| Terminal-Bench | 39.6% | Not measured | Only DeepSeek V3.2 measured |
| FrontierCode | Not measured | 43.9% | Only DeepSeek V4.1 Flash measured |
| ALE-Bench | Not measured | 1092.3 | Only DeepSeek V4.1 Flash measured |
| LMArena Text: coding | 1453 | 1502 | DeepSeek V4.1 Flash |
| LMArena Text: software and IT | 1449 | 1496 | DeepSeek V4.1 Flash |
| LMArena WebDev: React | 1350 | 1628 | DeepSeek V4.1 Flash |
| LMArena WebDev: HTML | 1371 | 1612 | DeepSeek V4.1 Flash |
| LMArena WebDev: from an image | Not measured | 1606 | Only DeepSeek V4.1 Flash measured |
| Neither model has a result on the other 11 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | Leader |
|---|---|---|---|
| AA Intelligence Index | 21.5 | 24.7 | DeepSeek V4.1 Flash |
| Epoch Capabilities Index | 146.1 | 155.9 | DeepSeek V4.1 Flash |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | Leader |
|---|---|---|---|
| LiveBench research | Not measured | 86.8% | Only DeepSeek V4.1 Flash measured |
| LiveBench browser | Not measured | 84.0% | Only DeepSeek V4.1 Flash measured |
| LiveBench long agentic | Not measured | 40.1% | Only DeepSeek V4.1 Flash measured |
| SREBench (Vals AI) | Not measured | 0.8% | Only DeepSeek V4.1 Flash measured |
| Finance Agent (Vals AI) | Not measured | 53.5% | Only DeepSeek V4.1 Flash measured |
| Tax Agent Bench (Vals AI) | Not measured | 26.6% | Only DeepSeek V4.1 Flash measured |
| Legal Research Bench (Vals AI) | Not measured | 41.4% | Only DeepSeek V4.1 Flash measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 57.2% | Only DeepSeek V4.1 Flash measured |
| BioMysteryBench (Vals AI) | Not measured | 67.8% | Only DeepSeek V4.1 Flash measured |
| Terminal-Bench-Science (Vals AI) | Not measured | 0.0% | Only DeepSeek V4.1 Flash measured |
| Vals Science Index | Not measured | 29.6% | Only DeepSeek V4.1 Flash measured |
| τ²-Bench Telecom | 90.6% | Not measured | Only DeepSeek V3.2 measured |
| APEX-Agents | 21.3% | 39.5% | DeepSeek V4.1 Flash |
| Vending-Bench 2 | $1,034 | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 27 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | Leader |
|---|---|---|---|
| GPQA Diamond | 80.3% | Not measured | Only DeepSeek V3.2 measured |
| LiveBench reasoning | Not measured | 86.7% | Only DeepSeek V4.1 Flash measured |
| MysteryMechanism (Vals AI) | Not measured | 21.2% | Only DeepSeek V4.1 Flash measured |
| GPQA Diamond (Artificial Analysis run) | 84.0% | Not measured | Only DeepSeek V3.2 measured |
| Humanity's Last Exam (Artificial Analysis run) | 24.6% | 39.2% | DeepSeek V4.1 Flash |
| GPQA Diamond (Epoch AI run) | Not measured | 89.8% | Only DeepSeek V4.1 Flash measured |
| ARC-AGI | 57.0% | 94.5% | DeepSeek V4.1 Flash |
| ARC-AGI-2 | 4.0% | 72.9% | DeepSeek V4.1 Flash |
| Chess puzzles | Not measured | 48.0% | Only DeepSeek V4.1 Flash measured |
| Mystery game puzzles | Not measured | 43.0% | Only DeepSeek V4.1 Flash measured |
| LMArena Text: hard prompts | 1434 | 1483 | DeepSeek V4.1 Flash |
| LMArena Text: expert | 1436 | 1501 | DeepSeek V4.1 Flash |
| LMArena Text: hard prompts in English | 1446 | 1488 | DeepSeek V4.1 Flash |
| LMArena Vision: diagrams | Not measured | 1297 | Only DeepSeek V4.1 Flash measured |
| LMArena Vision: homework | Not measured | 1307 | Only DeepSeek V4.1 Flash measured |
| Neither model has a result on the other 7 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | 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 | 98.3% | Only DeepSeek V4.1 Flash measured |
| FrontierMath, tiers 1 to 3 | Not measured | 67.4% | Only DeepSeek V4.1 Flash measured |
| FrontierMath, tier 4 | Not measured | 26.8% | Only DeepSeek V4.1 Flash measured |
| ProofBench | Not measured | 54.0% | Only DeepSeek V4.1 Flash measured |
| LMArena Text: math | 1435 | 1478 | DeepSeek V4.1 Flash |
| LMArena Text: mathematical work | 1440 | 1485 | DeepSeek V4.1 Flash |
| Neither model has a result on the other 3 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index05
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | Leader |
|---|---|---|---|
| MMLU Pro | 84.9% | Not measured | Only DeepSeek V3.2 measured |
| LegalBench (Vals AI) | 76.1% | 83.3% | DeepSeek V4.1 Flash |
| CaseLaw (Vals AI) | 55.4% | Not measured | Only DeepSeek V3.2 measured |
| CorpFin (Vals AI) | 51.0% | Not measured | Only DeepSeek V3.2 measured |
| TaxEval (Vals AI) | 68.2% | Not measured | Only DeepSeek V3.2 measured |
| MedQA (Vals AI) | 93.9% | Not measured | Only DeepSeek V3.2 measured |
| MedCode (Vals AI) | Not measured | 41.2% | Only DeepSeek V4.1 Flash measured |
| MMLU Pro (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| SimpleQA Verified | Not measured | 42.9% | Only DeepSeek V4.1 Flash measured |
| LMArena Text: science | 1446 | 1469 | DeepSeek V4.1 Flash |
| LMArena Text: medicine and healthcare | 1440 | 1484 | DeepSeek V4.1 Flash |
| LMArena Text: legal and government | 1425 | 1443 | DeepSeek V4.1 Flash |
| LMArena Text: business and finance | 1410 | 1461 | DeepSeek V4.1 Flash |
| Neither model has a result on the other 5 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index08
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | Leader |
|---|---|---|---|
| Design Arena: websites | 1175 | 1308 | DeepSeek V4.1 Flash |
| Design Arena: UI components | 1146 | 1312 | DeepSeek V4.1 Flash |
| Design Arena: game dev | 1131 | 1330 | DeepSeek V4.1 Flash |
| Design Arena: data viz | 1158 | 1291 | DeepSeek V4.1 Flash |
| Design Arena: 3D | 1132 | 1355 | DeepSeek V4.1 Flash |
| Design Arena: all code categories | 1163 | 1320 | DeepSeek V4.1 Flash |
| Design Arena: SVG | 1032 | 1245 | DeepSeek V4.1 Flash |
| Design Arena: ASCII art | 1083 | 1265 | DeepSeek V4.1 Flash |
| LiveBench design | Not measured | 79.2% | Only DeepSeek V4.1 Flash measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index020
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | Leader |
|---|---|---|---|
| LiveBench data analysis | Not measured | 79.3% | Only DeepSeek V4.1 Flash measured |
| Public Benefits Bench (Vals AI) | Not measured | 64.3% | Only DeepSeek V4.1 Flash measured |
| SAGE (Vals AI) | Not measured | 47.9% | Only DeepSeek V4.1 Flash measured |
| MedScribe (Vals AI) | Not measured | 85.5% | Only DeepSeek V4.1 Flash measured |
| IFBench | 60.7% | Not measured | Only DeepSeek V3.2 measured |
| Long context reasoning (AA-LCR) | 73.3% | 84.0% | DeepSeek V4.1 Flash |
| DTBench | 85.6% | 89.9% | DeepSeek V4.1 Flash |
| LMCA | 28.8% | 47.0% | DeepSeek V4.1 Flash |
| CL-bench | 12.4% | Not measured | Only DeepSeek V3.2 measured |
| Surface Evolver Bench | Not measured | 46.3% | Only DeepSeek V4.1 Flash measured |
| Furniture assembly | Not measured | 34.2% | Only DeepSeek V4.1 Flash measured |
| CL-bench Life | 7.4% | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text | 1425 | 1464 | DeepSeek V4.1 Flash |
| LMArena Text: instruction following | 1413 | 1473 | DeepSeek V4.1 Flash |
| LMArena Text: creative writing | 1400 | 1437 | DeepSeek V4.1 Flash |
| LMArena Vision | Not measured | 1278 | Only DeepSeek V4.1 Flash measured |
| LMArena Text: English | 1439 | 1473 | DeepSeek V4.1 Flash |
| LMArena Text: non-English | 1409 | 1449 | DeepSeek V4.1 Flash |
| LMArena Text: Chinese | 1455 | 1501 | DeepSeek V4.1 Flash |
| LMArena Text: German | 1413 | 1470 | DeepSeek V4.1 Flash |
| LMArena Text: Japanese | 1374 | 1432 | DeepSeek V4.1 Flash |
| LMArena Text: Korean | 1370 | 1436 | DeepSeek V4.1 Flash |
| LMArena Text: French | 1429 | 1465 | DeepSeek V4.1 Flash |
| LMArena Text: Spanish | 1424 | 1459 | DeepSeek V4.1 Flash |
| LMArena Text: Russian | 1424 | 1467 | DeepSeek V4.1 Flash |
| LMArena Text: Polish | 1419 | 1457 | DeepSeek V4.1 Flash |
| LMArena Text: multi-turn | 1427 | 1461 | DeepSeek V4.1 Flash |
| LMArena Text: longer queries | 1428 | 1473 | DeepSeek V4.1 Flash |
| LMArena Text: writing and language | 1407 | 1450 | DeepSeek V4.1 Flash |
| LMArena Text: entertainment and media | 1395 | 1445 | DeepSeek V4.1 Flash |
| LMArena Vision: English | Not measured | 1288 | Only DeepSeek V4.1 Flash measured |
| LMArena Vision: Chinese | Not measured | 1327 | Only DeepSeek V4.1 Flash measured |
| LMArena Vision: reading text | Not measured | 1282 | Only DeepSeek V4.1 Flash measured |
| LMArena Vision: humor | Not measured | 1260 | Only DeepSeek V4.1 Flash measured |
| LMArena Vision: creative writing | Not measured | 1275 | Only DeepSeek V4.1 Flash measured |
| Neither model has a result on the other 11 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | DeepSeek V3.2 | DeepSeek V4.1 Flash | 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$0.0007Fits in one request
DeepSeek V4.1 Flash$0.0006Fits in one request
DeepSeek V4.1 Flash costs less
Repository review
50K input + 3K output tokens
DeepSeek V3.2$0.017Fits in one request
DeepSeek V4.1 Flash$0.014Fits in one request
DeepSeek V4.1 Flash costs less
Agent loop
200K input + 10K output tokens
DeepSeek V3.2$0.067Exceeds the context window
DeepSeek V4.1 Flash$0.053Fits in one request
DeepSeek V4.1 Flash 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 | DeepSeek V4.1 Flash |
|---|---|---|
| Maker | DeepSeek | DeepSeek |
| Weights | Open weights | Open weights |
| Released | December 1, 2025 | September 10, 2026 |
| Context window | 164K | 1.05M |
| Input $/Mtok | $0.29 | $0.2225 |
| Output $/Mtok | $0.88 | $0.84 |
| Coding index | 30.0 | 62.5 |
| Independent measurements behind it | 3 | 5 |
| Cost per solved task | $0.99 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is DeepSeek V3.2 better than DeepSeek V4.1 Flash for coding?
Pick DeepSeek V4.1 Flash for coding: it scores 62.5 on the Vibe Coding Index against 30.0 for DeepSeek V3.2. DeepSeek V3.2 is neither cheaper nor larger in context, and leads no benchmark both models were measured on.
Which is cheaper, DeepSeek V3.2 or DeepSeek V4.1 Flash?
DeepSeek V3.2 costs $0.29 per million input tokens and $0.88 per million output tokens; DeepSeek V4.1 Flash costs $0.2225 and $0.84. Cost per solved task is known for DeepSeek V3.2 only.
Which has the bigger context window, DeepSeek V3.2 or DeepSeek V4.1 Flash?
DeepSeek V4.1 Flash, with 1.05M tokens against 164K.
How many benchmarks is the DeepSeek V3.2 vs DeepSeek V4.1 Flash comparison based on?
50 benchmarks have a result for both models as of October 22, 2026. DeepSeek V3.2's coding index rests on 3 independent benchmarks and DeepSeek V4.1 Flash's on 5. Where only one model has a result, the page says so instead of naming a winner.