DeepSeek V3.2 vs Gemini 4 Argon
Gemini 4 Argon leads DeepSeek V3.2 on coding, 81.0 to 30.0 on the Vibe Coding Index, and DeepSeek V3.2 costs 11.4x less per output token ($0.88 vs $10.00 per million). Updated October 22, 2026; 44 benchmarks have a result for both.
DeepSeek V3.2 or Gemini 4 Argon: which to pick
Pick Gemini 4 Argon for coding: it scores 81.0 on the Vibe Coding Index against 30.0 for DeepSeek V3.2.
Pick DeepSeek V3.2 for output tokens 11.4x cheaper ($0.88 vs $10.00 per million).
- Shared results
- 44
- DeepSeek V3.2 only
- 31
- Gemini 4 Argon only
- 50
- 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
Gemini 4 Argon
Gemini 4 Argon scores 81.0 against 30.0.
3 and 5 benchmarks behind the scores
Intelligence
AA Intelligence Index
Gemini 4 Argon
Gemini 4 Argon leads, 52.6 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
Gemini 4 Argon
Gemini 4 Argon leads, 1315 against 1175.
Same board, both measured
General
Vals Index
No call
Only Gemini 4 Argon has a Vals Index result, 69.2%.
Directional only
Long documents
Listed context window
Gemini 4 Argon
Gemini 4 Argon takes 1.05M tokens against 164K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
DeepSeek V3.2
DeepSeek V3.2 $0.0007, Gemini 4 Argon $0.0070 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, Gemini 4 Argon 81.0 from 5 benchmarks.
- Widest category
- Agentic work, 70.2 points apart on average over 2 shared benchmarks.
- Widest single gap
- ARC-AGI-2: DeepSeek V3.2 4.0% against Gemini 4 Argon 91.3%, 87.2 points apart on a 0 to 100 scale.
- Output price
- DeepSeek V3.2 $0.88 against Gemini 4 Argon $10.00 per million output tokens, 11.4x apart.
- Context window
- DeepSeek V3.2 takes 164K tokens, Gemini 4 Argon 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 | Gemini 4 Argon | Ties |
|---|---|---|---|---|
| Coding | 0 | 6 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 2 | ||
| Reasoning | 0 | 6 | ||
| Math | 0 | 2 | ||
| Knowledge | 0 | 5 | ||
| Web design | 0 | 6 | ||
| General | 0 | 15 | ||
| All shared boards | 0 | 44 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index06
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | 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 | 89.7% | Only Gemini 4 Argon measured |
| LMArena WebDev | 1362 | 1678 | Gemini 4 Argon |
| 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 | 57.6% | Only Gemini 4 Argon measured |
| Vibe Code Bench (Vals AI) | 5.1% | 91.9% | Gemini 4 Argon |
| Code migration (Vals AI) | Not measured | 68.2% | Only Gemini 4 Argon measured |
| ProgramBench (Vals AI) | Not measured | 2.5% | Only Gemini 4 Argon measured |
| IOI (Vals AI) | Not measured | 100.0% | Only Gemini 4 Argon measured |
| Terminal-Bench 2.1 (Vals AI) | Not measured | 57.7% | Only Gemini 4 Argon measured |
| SWE-bench Multimodal (Vals AI) | Not measured | 56.0% | Only Gemini 4 Argon measured |
| Vibe Code Bench 1-100 (Vals AI) | Not measured | 21.2% | Only Gemini 4 Argon measured |
| CyberBench (Vals AI) | Not measured | 77.9% | Only Gemini 4 Argon measured |
| LiveCodeBench (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| SciCode | Not measured | 61.8% | Only Gemini 4 Argon 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 | 47.5% | Only Gemini 4 Argon measured |
| FrontierSWE | Not measured | 55.0% | Only Gemini 4 Argon measured |
| MirrorCode | Not measured | 58.9% | Only Gemini 4 Argon measured |
| ALE-Bench | Not measured | 2100.1 | Only Gemini 4 Argon measured |
| LMArena Text: coding | 1453 | 1546 | Gemini 4 Argon |
| LMArena Text: software and IT | 1449 | 1556 | Gemini 4 Argon |
| LMArena WebDev: React | 1350 | 1692 | Gemini 4 Argon |
| LMArena WebDev: HTML | 1371 | 1643 | Gemini 4 Argon |
| SciCode | Not measured | 12.3% | Only Gemini 4 Argon measured |
| Neither model has a result on the other 10 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | Leader |
|---|---|---|---|
| AA Intelligence Index | 21.5 | 52.6 | Gemini 4 Argon |
| Epoch Capabilities Index | 146.1 | 166.4 | Gemini 4 Argon |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | Leader |
|---|---|---|---|
| Design Arena: spreadsheets | Not measured | 1157 | Only Gemini 4 Argon measured |
| LiveBench research | Not measured | 83.1% | Only Gemini 4 Argon measured |
| LiveBench browser | Not measured | 92.0% | Only Gemini 4 Argon measured |
| LiveBench long agentic | Not measured | 69.9% | Only Gemini 4 Argon measured |
| SREBench (Vals AI) | Not measured | 44.3% | Only Gemini 4 Argon measured |
| Finance Agent (Vals AI) | Not measured | 65.4% | Only Gemini 4 Argon measured |
| Tax Agent Bench (Vals AI) | Not measured | 44.6% | Only Gemini 4 Argon measured |
| Legal Research Bench (Vals AI) | Not measured | 54.8% | Only Gemini 4 Argon measured |
| Harvey's Legal Agent Benchmark (Vals AI) | Not measured | 27.1% | Only Gemini 4 Argon measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 75.2% | Only Gemini 4 Argon measured |
| BioMysteryBench (Vals AI) | Not measured | 76.3% | Only Gemini 4 Argon measured |
| Terminal-Bench-Science (Vals AI) | Not measured | 44.3% | Only Gemini 4 Argon measured |
| Vals Science Index | Not measured | 55.4% | Only Gemini 4 Argon measured |
| Time Horizon Index (Vals AI) | Not measured | 52.5% | Only Gemini 4 Argon measured |
| CUA-bench (Vals AI) | Not measured | 4.8% | Only Gemini 4 Argon measured |
| Robo-bench (Vals AI) | Not measured | 0.8% | Only Gemini 4 Argon measured |
| τ²-Bench Telecom | 90.6% | Not measured | Only DeepSeek V3.2 measured |
| APEX-Agents | 21.3% | 82.4% | Gemini 4 Argon |
| Vending-Bench 2 | $1,034 | $13,718 | Gemini 4 Argon |
| Neither model has a result on the other 22 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | Leader |
|---|---|---|---|
| GPQA Diamond | 80.3% | Not measured | Only DeepSeek V3.2 measured |
| LiveBench reasoning | Not measured | 95.1% | Only Gemini 4 Argon measured |
| MysteryMechanism (Vals AI) | Not measured | 45.5% | Only Gemini 4 Argon measured |
| GPQA Diamond (Artificial Analysis run) | 84.0% | Not measured | Only DeepSeek V3.2 measured |
| Humanity's Last Exam (Artificial Analysis run) | 24.6% | 57.1% | Gemini 4 Argon |
| GPQA Diamond (Epoch AI run) | Not measured | 96.4% | Only Gemini 4 Argon measured |
| ARC-AGI | 57.0% | 98.5% | Gemini 4 Argon |
| ARC-AGI-2 | 4.0% | 91.3% | Gemini 4 Argon |
| Chess puzzles | Not measured | 59.0% | Only Gemini 4 Argon measured |
| Mystery game puzzles | Not measured | 64.0% | Only Gemini 4 Argon measured |
| EBR-bench | Not measured | 64.8% | Only Gemini 4 Argon measured |
| LMArena Text: hard prompts | 1434 | 1546 | Gemini 4 Argon |
| LMArena Text: expert | 1436 | 1550 | Gemini 4 Argon |
| LMArena Text: hard prompts in English | 1446 | 1548 | Gemini 4 Argon |
| Neither model has a result on the other 8 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | 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 | 100.0% | Only Gemini 4 Argon measured |
| FrontierMath, tiers 1 to 3 | Not measured | 86.0% | Only Gemini 4 Argon measured |
| FrontierMath, tier 4 | Not measured | 78.0% | Only Gemini 4 Argon measured |
| ProofBench | Not measured | 99.0% | Only Gemini 4 Argon measured |
| LMArena Text: math | 1435 | 1536 | Gemini 4 Argon |
| LMArena Text: mathematical work | 1440 | 1557 | Gemini 4 Argon |
| 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 | Gemini 4 Argon | Leader |
|---|---|---|---|
| MMLU Pro | 84.9% | Not measured | Only DeepSeek V3.2 measured |
| LegalBench (Vals AI) | 76.1% | 88.3% | Gemini 4 Argon |
| 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 | 58.8% | Only Gemini 4 Argon measured |
| MMLU Pro (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| SimpleQA Verified | Not measured | 76.0% | Only Gemini 4 Argon measured |
| LMArena Text: science | 1446 | 1548 | Gemini 4 Argon |
| LMArena Text: medicine and healthcare | 1440 | 1555 | Gemini 4 Argon |
| LMArena Text: legal and government | 1425 | 1548 | Gemini 4 Argon |
| LMArena Text: business and finance | 1410 | 1525 | Gemini 4 Argon |
| Neither model has a result on the other 5 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | Leader |
|---|---|---|---|
| Design Arena: websites | 1175 | 1315 | Gemini 4 Argon |
| Design Arena: UI components | 1146 | 1333 | Gemini 4 Argon |
| Design Arena: game dev | 1131 | 1377 | Gemini 4 Argon |
| Design Arena: data viz | 1158 | 1332 | Gemini 4 Argon |
| Design Arena: 3D | 1132 | 1388 | Gemini 4 Argon |
| Design Arena: all code categories | 1163 | 1329 | Gemini 4 Argon |
| Design Arena: SVG | 1032 | Not measured | Only DeepSeek V3.2 measured |
| Design Arena: ASCII art | 1083 | Not measured | Only DeepSeek V3.2 measured |
| LiveBench design | Not measured | 73.7% | Only Gemini 4 Argon measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index015
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | Leader |
|---|---|---|---|
| Vals Index | Not measured | 69.2% | Only Gemini 4 Argon measured |
| LiveBench data analysis | Not measured | 85.2% | Only Gemini 4 Argon measured |
| Public Benefits Bench (Vals AI) | Not measured | 69.8% | Only Gemini 4 Argon measured |
| SAGE (Vals AI) | Not measured | 53.6% | Only Gemini 4 Argon measured |
| MedScribe (Vals AI) | Not measured | 87.4% | Only Gemini 4 Argon measured |
| IFBench | 60.7% | Not measured | Only DeepSeek V3.2 measured |
| Long context reasoning (AA-LCR) | 73.3% | 79.7% | Gemini 4 Argon |
| DTBench | 85.6% | 97.6% | Gemini 4 Argon |
| LMCA | 28.8% | 63.7% | Gemini 4 Argon |
| CL-bench | 12.4% | Not measured | Only DeepSeek V3.2 measured |
| Furniture assembly | Not measured | 72.5% | Only Gemini 4 Argon measured |
| CL-bench Life | 7.4% | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text | 1425 | 1532 | Gemini 4 Argon |
| LMArena Text: instruction following | 1413 | 1534 | Gemini 4 Argon |
| LMArena Text: creative writing | 1400 | 1527 | Gemini 4 Argon |
| LMArena Text: English | 1439 | 1538 | Gemini 4 Argon |
| LMArena Text: non-English | 1409 | 1522 | Gemini 4 Argon |
| LMArena Text: Chinese | 1455 | 1611 | Gemini 4 Argon |
| LMArena Text: German | 1413 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: Japanese | 1374 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: Korean | 1370 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: French | 1429 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: Spanish | 1424 | 1503 | Gemini 4 Argon |
| LMArena Text: Russian | 1424 | 1534 | Gemini 4 Argon |
| LMArena Text: Polish | 1419 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: multi-turn | 1427 | 1556 | Gemini 4 Argon |
| LMArena Text: longer queries | 1428 | 1547 | Gemini 4 Argon |
| LMArena Text: writing and language | 1407 | 1529 | Gemini 4 Argon |
| LMArena Text: entertainment and media | 1395 | 1527 | Gemini 4 Argon |
| Neither model has a result on the other 17 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | DeepSeek V3.2 | Gemini 4 Argon | 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 | Google Gemini 4 Argon |
|---|---|---|
| Maker | DeepSeek | |
| Weights | Open weights | No public weights listed |
| Released | December 1, 2025 | October 13, 2026 |
| Context window | 164K | 1.05M |
| Input $/Mtok | $0.29 | $2.00 |
| Output $/Mtok | $0.88 | $10.00 |
| Coding index | 30.0 | 81.0 |
| 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 Gemini 4 Argon for coding?
Pick Gemini 4 Argon for coding: it scores 81.0 on the Vibe Coding Index against 30.0 for DeepSeek V3.2. Pick DeepSeek V3.2 for output tokens 11.4x cheaper ($0.88 vs $10.00 per million).
Which is cheaper, DeepSeek V3.2 or Gemini 4 Argon?
DeepSeek V3.2 costs $0.29 per million input tokens and $0.88 per million output tokens; Gemini 4 Argon costs $2.00 and $10.00. Cost per solved task is known for DeepSeek V3.2 only.
Which has the bigger context window, DeepSeek V3.2 or Gemini 4 Argon?
Gemini 4 Argon, with 1.05M tokens against 164K.
How many benchmarks is the DeepSeek V3.2 vs Gemini 4 Argon comparison based on?
44 benchmarks have a result for both models as of October 22, 2026. DeepSeek V3.2's coding index rests on 3 independent benchmarks and Gemini 4 Argon's on 5. Where only one model has a result, the page says so instead of naming a winner.