DeepSeek V3.2 vs Grok 4.20
Grok 4.20 leads DeepSeek V3.2 on coding, 33.3 to 30.0 on the Vibe Coding Index, and DeepSeek V3.2 costs 2.8x less per output token ($0.88 vs $2.50 per million). Updated October 21, 2026; 35 benchmarks have a result for both.
DeepSeek V3.2 or Grok 4.20: which to pick
Pick Grok 4.20 for coding: it scores 33.3 on the Vibe Coding Index against 30.0 for DeepSeek V3.2, though on 2 benchmarks against 3.
Pick DeepSeek V3.2 for output tokens 2.8x cheaper ($0.88 vs $2.50 per million) and work measured by Vibe Code Bench (Vals AI) or CaseLaw (Vals AI), where it leads.
- Shared results
- 35
- DeepSeek V3.2 only
- 40
- Grok 4.20 only
- 19
- 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
Grok 4.20
Grok 4.20 scores 33.3 against 30.0.
3 and 2 benchmarks behind the scores
Intelligence
AA Intelligence Index
Grok 4.20
Grok 4.20 leads, 25.7 against 21.5.
Same board, both measured
Reasoning
GPQA Diamond
Grok 4.20
Grok 4.20 leads, 88.6% against 80.3%.
Same board, both measured
Math
AIME
Grok 4.20
Grok 4.20 leads, 96.5% against 84.6%.
Same board, both measured
Knowledge
MMLU Pro
Grok 4.20
Grok 4.20 leads, 86.3% against 84.9%.
Same board, both measured
Web design
Design Arena: websites
Grok 4.20
Grok 4.20 leads, 1229 against 1175.
Same board, both measured
Long documents
Listed context window
Grok 4.20
Grok 4.20 takes 2M tokens against 164K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
DeepSeek V3.2
DeepSeek V3.2 $0.0007, Grok 4.20 $0.0025 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Grok 4.20
DeepSeek V3.2 $0.99, Grok 4.20 $0.72.
Published attempt cost and pass rate
At a glance
- Coding index
- DeepSeek V3.2 30.0 from 3 benchmarks, Grok 4.20 33.3 from 2 benchmarks.
- Widest category
- Reasoning, 23.8 points apart on average over 5 shared benchmarks.
- Widest single gap
- ARC-AGI-2: DeepSeek V3.2 4.0% against Grok 4.20 65.1%, 61.1 points apart on a 0 to 100 scale.
- Output price
- DeepSeek V3.2 $0.88 against Grok 4.20 $2.50 per million output tokens, 2.8x apart.
- Context window
- DeepSeek V3.2 takes 164K tokens, Grok 4.20 2M.
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 | Grok 4.20 | Ties |
|---|---|---|---|---|
| Coding | 1 | 4 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 2 | ||
| Reasoning | 0 | 5 | ||
| Math | 0 | 1 | ||
| Knowledge | 1 | 5 | ||
| Web design | 0 | 8 | ||
| General | 1 | 5 | ||
| All shared boards | 3 | 32 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index14
| Benchmark | DeepSeek V3.2 | Grok 4.20 | 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 |
| LMArena WebDev | 1362 | Not measured | Only DeepSeek V3.2 measured |
| AA Coding Index | 44.2 | Not measured | Only DeepSeek V3.2 measured |
| LiveCodeBench (Vals AI) | 80.7% | 84.3% | Grok 4.20 |
| SWE-bench Verified (Vals AI) | 67.6% | 72.2% | Grok 4.20 |
| Vibe Code Bench (Vals AI) | 5.1% | 4.1% | DeepSeek V3.2 |
| Code migration (Vals AI) | Not measured | 0.3% | Only Grok 4.20 measured |
| Terminal-Bench 2.1 (Vals AI) | Not measured | 44.2% | Only Grok 4.20 measured |
| LiveCodeBench (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 35.6% | 37.9% | Grok 4.20 |
| Terminal-Bench | 39.6% | 57.3% | Grok 4.20 |
| WeirdML | Not measured | 52.3% | Only Grok 4.20 measured |
| LMArena Text: coding | 1453 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: software and IT | 1449 | Not measured | Only DeepSeek V3.2 measured |
| LMArena WebDev: React | 1350 | Not measured | Only DeepSeek V3.2 measured |
| LMArena WebDev: HTML | 1371 | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 21 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | Grok 4.20 | Leader |
|---|---|---|---|
| AA Intelligence Index | 21.5 | 25.7 | Grok 4.20 |
| Epoch Capabilities Index | 146.1 | 152.1 | Grok 4.20 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V3.2 | Grok 4.20 | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | Not measured | 1144 | Only Grok 4.20 measured |
| Design Arena: agents building full-stack apps | Not measured | 1054 | Only Grok 4.20 measured |
| Design Arena: agents building mobile apps | Not measured | 1090 | Only Grok 4.20 measured |
| Design Arena: agents building Android apps | Not measured | 1052 | Only Grok 4.20 measured |
| Design Arena: agents building Godot games | Not measured | 1089 | Only Grok 4.20 measured |
| Design Arena: HTML slides | Not measured | 1180 | Only Grok 4.20 measured |
| Finance Agent (Vals AI) | Not measured | 28.5% | Only Grok 4.20 measured |
| Tax Agent Bench (Vals AI) | Not measured | 7.0% | Only Grok 4.20 measured |
| Legal Research Bench (Vals AI) | Not measured | 13.9% | Only Grok 4.20 measured |
| Excel Modeling Benchmark (Vals AI) | Not measured | 11.7% | Only Grok 4.20 measured |
| τ²-Bench Telecom | 90.6% | 93.0% | Grok 4.20 |
| APEX-Agents | 21.3% | Not measured | Only DeepSeek V3.2 measured |
| Vending-Bench 2 | $1,034 | $4,663 | Grok 4.20 |
| Neither model has a result on the other 26 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index05
| Benchmark | DeepSeek V3.2 | Grok 4.20 | Leader |
|---|---|---|---|
| GPQA Diamond | 80.3% | 88.6% | Grok 4.20 |
| MMMU Pro (Vals AI) | Not measured | 83.5% | Only Grok 4.20 measured |
| GPQA Diamond (Artificial Analysis run) | 84.0% | 91.1% | Grok 4.20 |
| Humanity's Last Exam (Artificial Analysis run) | 24.6% | 34.5% | Grok 4.20 |
| ARC-AGI | 57.0% | 89.5% | Grok 4.20 |
| ARC-AGI-2 | 4.0% | 65.1% | Grok 4.20 |
| LMArena Text: hard prompts | 1434 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: expert | 1436 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: hard prompts in English | 1446 | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 12 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | DeepSeek V3.2 | Grok 4.20 | Leader |
|---|---|---|---|
| AIME | 84.6% | 96.5% | Grok 4.20 |
| 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 |
| LMArena Text: math | 1435 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: mathematical work | 1440 | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index15
| Benchmark | DeepSeek V3.2 | Grok 4.20 | Leader |
|---|---|---|---|
| MMLU Pro | 84.9% | 86.3% | Grok 4.20 |
| LegalBench (Vals AI) | 76.1% | 77.7% | Grok 4.20 |
| CaseLaw (Vals AI) | 55.4% | 54.4% | DeepSeek V3.2 |
| CorpFin (Vals AI) | 51.0% | 63.7% | Grok 4.20 |
| TaxEval (Vals AI) | 68.2% | 74.1% | Grok 4.20 |
| MortgageTax (Vals AI) | Not measured | 45.4% | Only Grok 4.20 measured |
| MedQA (Vals AI) | 93.9% | 94.5% | Grok 4.20 |
| MedCode (Vals AI) | Not measured | 32.2% | Only Grok 4.20 measured |
| MMLU Pro (Artificial Analysis run) | 86.2% | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: science | 1446 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: medicine and healthcare | 1440 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: legal and government | 1425 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: business and finance | 1410 | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 4 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index08
| Benchmark | DeepSeek V3.2 | Grok 4.20 | Leader |
|---|---|---|---|
| Design Arena: websites | 1175 | 1229 | Grok 4.20 |
| Design Arena: UI components | 1149 | 1199 | Grok 4.20 |
| Design Arena: game dev | 1132 | 1192 | Grok 4.20 |
| Design Arena: data viz | 1158 | 1217 | Grok 4.20 |
| Design Arena: 3D | 1133 | 1193 | Grok 4.20 |
| Design Arena: all code categories | 1164 | 1221 | Grok 4.20 |
| Design Arena: SVG | 1032 | 1164 | Grok 4.20 |
| Design Arena: ASCII art | 1083 | 1174 | Grok 4.20 |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index15
| Benchmark | DeepSeek V3.2 | Grok 4.20 | Leader |
|---|---|---|---|
| SAGE (Vals AI) | Not measured | 38.2% | Only Grok 4.20 measured |
| MedScribe (Vals AI) | Not measured | 63.4% | Only Grok 4.20 measured |
| Vals Multimodal Index | Not measured | 39.1% | Only Grok 4.20 measured |
| IFBench | 60.7% | 81.2% | Grok 4.20 |
| Long context reasoning (AA-LCR) | 73.3% | 69.0% | DeepSeek V3.2 |
| DTBench | 85.6% | 90.1% | Grok 4.20 |
| LMCA | 28.8% | 38.7% | Grok 4.20 |
| CL-bench | 12.4% | 22.2% | Grok 4.20 |
| CL-bench Life | 7.4% | 11.9% | Grok 4.20 |
| LMArena Text | 1425 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: instruction following | 1413 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: creative writing | 1400 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: English | 1439 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: non-English | 1409 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: Chinese | 1455 | Not measured | Only DeepSeek V3.2 measured |
| 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 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: Russian | 1424 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: Polish | 1419 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: multi-turn | 1427 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: longer queries | 1428 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: writing and language | 1407 | Not measured | Only DeepSeek V3.2 measured |
| LMArena Text: entertainment and media | 1395 | Not measured | Only DeepSeek V3.2 measured |
| Neither model has a result on the other 20 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | DeepSeek V3.2 | Grok 4.20 | 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 | SpaceXAI Grok 4.20 |
|---|---|---|
| Maker | DeepSeek | xAI |
| Weights | Open weights | No public weights listed |
| Released | December 1, 2025 | March 31, 2026 |
| Context window | 164K | 2M |
| Input $/Mtok | $0.29 | $1.25 |
| Output $/Mtok | $0.88 | $2.50 |
| Coding index | 30.0 | 33.3 |
| Independent measurements behind it | 3 | 2 |
| Cost per solved task | $0.99 | $0.72 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is DeepSeek V3.2 better than Grok 4.20 for coding?
Pick Grok 4.20 for coding: it scores 33.3 on the Vibe Coding Index against 30.0 for DeepSeek V3.2, though on 2 benchmarks against 3. Pick DeepSeek V3.2 for output tokens 2.8x cheaper ($0.88 vs $2.50 per million) and work measured by Vibe Code Bench (Vals AI) or CaseLaw (Vals AI), where it leads.
Which is cheaper, DeepSeek V3.2 or Grok 4.20?
DeepSeek V3.2 costs $0.29 per million input tokens and $0.88 per million output tokens; Grok 4.20 costs $1.25 and $2.50. Per solved SWE-bench Verified task, DeepSeek V3.2 costs $0.99 and Grok 4.20 $0.72.
Which has the bigger context window, DeepSeek V3.2 or Grok 4.20?
Grok 4.20, with 2M tokens against 164K.
How many benchmarks is the DeepSeek V3.2 vs Grok 4.20 comparison based on?
35 benchmarks have a result for both models as of October 21, 2026. DeepSeek V3.2's coding index rests on 3 independent benchmarks and Grok 4.20's on 2. Where only one model has a result, the page says so instead of naming a winner.