DeepSeek V4 Pro 0423 vs MiniMax M3
Updated October 21, 2026. 75 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Pick MiniMax M3 for coding: it scores 57.8 on the Vibe Coding Index against 35.3 for DeepSeek V4 Pro 0423.
Pick DeepSeek V4 Pro 0423 when what matters is AA Coding Index or LiveCodeBench (Vals AI), where it leads.
- Shared results
- 75
- DeepSeek V4 Pro 0423 only
- 8
- MiniMax M3 only
- 41
- Like-for-like categories
- 8 of 9
Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.
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, four public coding benchmarks
MiniMax M3
MiniMax M3 scores 57.8 against 35.3.
1 and 2 benchmarks behind the scores
Intelligence
AA Intelligence Index
DeepSeek V4 Pro 0423
DeepSeek V4 Pro 0423 leads, 30.4 against 29.2.
Same board, both measured
Agentic work
AA Agentic Index
MiniMax M3
MiniMax M3 leads, 29.5 against 26.3.
Same board, both measured
Reasoning
GPQA Diamond
MiniMax M3
MiniMax M3 leads, 92.7% against 89.4%.
Same board, both measured
Knowledge
MMLU Pro
DeepSeek V4 Pro 0423
DeepSeek V4 Pro 0423 leads, 87.3% against 84.2%.
Same board, both measured
Web design
Design Arena: websites
MiniMax M3
MiniMax M3 leads, 1257 against 1235.
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
MiniMax M3
DeepSeek V4 Pro 0423 $0.0019, MiniMax M3 $0.0009 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
MiniMax M3
DeepSeek V4 Pro 0423 $0.57, MiniMax M3 $0.56.
Published attempt cost and pass rate
At a glance
- Coding index
- DeepSeek V4 Pro 0423 35.3 from 1 benchmark, MiniMax M3 57.8 from 2 benchmarks.
- Widest category
- Agentic work, 10.7 points apart on average over 8 shared benchmarks.
- Widest single gap
- Design Arena: agents building full-stack apps: DeepSeek V4 Pro 0423 948 against MiniMax M3 1197, 31.1 points apart on a 0 to 100 scale.
- Output price
- DeepSeek V4 Pro 0423 $1.9105 against MiniMax M3 $1.20 per million output tokens, 1.6x 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 Pro 0423 | Bars | MiniMax M3 | Ties |
|---|---|---|---|---|
| Coding | 9 | 6 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 3 | 5 | ||
| Reasoning | 5 | 1 | 1 tie | |
| Math | 3 | 1 | ||
| Knowledge | 5 | 4 | ||
| Web design | 2 | 6 | ||
| General | 19 | 3 | ||
| All shared boards | 48 | 26 | 1 tie |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index96
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| LiveBench coding | Not measured | 70.9% | Only MiniMax M3 measured |
| LMArena WebDev | 1447 | 1482 | MiniMax M3 |
| AA Coding Index | 59.4 | 58.6 | DeepSeek V4 Pro 0423 |
| LiveCodeBench (Vals AI) | 87.5% | 82.2% | DeepSeek V4 Pro 0423 |
| SWE-bench Verified (Vals AI) | 77.4% | 75.0% | DeepSeek V4 Pro 0423 |
| Terminal-Bench 4 (Vals AI) | 11.1% | 1.0% | DeepSeek V4 Pro 0423 |
| Vibe Code Bench (Vals AI) | 49.9% | 47.6% | DeepSeek V4 Pro 0423 |
| Code migration (Vals AI) | 26.2% | 19.9% | DeepSeek V4 Pro 0423 |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| Terminal-Bench 2.1 (Vals AI) | 50.2% | 53.6% | MiniMax M3 |
| SWE-bench Multimodal (Vals AI) | Not measured | 33.3% | Only MiniMax M3 measured |
| Vibe Code Bench 1-100 (Vals AI) | Not measured | 9.2% | Only MiniMax M3 measured |
| SkillsBench (Vals AI) | 51.3% | 51.5% | MiniMax M3 |
| CyberBench (Vals AI) | Not measured | 65.0% | Only MiniMax M3 measured |
| SciCode | Not measured | 47.1% | Only MiniMax M3 measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 42.4% | Only MiniMax M3 measured |
| FrontierCode | 17.6% | 14.7% | DeepSeek V4 Pro 0423 |
| WeirdML | 48.9% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| ALE-Bench | 1006.1 | 640.0 | DeepSeek V4 Pro 0423 |
| LMArena Text: coding | 1470 | 1470 | MiniMax M3 |
| LMArena Text: software and IT | 1469 | 1466 | DeepSeek V4 Pro 0423 |
| LMArena WebDev: React | 1441 | 1478 | MiniMax M3 |
| LMArena WebDev: HTML | 1435 | 1455 | MiniMax M3 |
| LMArena WebDev: from an image | Not measured | 1470 | Only MiniMax M3 measured |
| Neither model has a result on the other 14 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| AA Intelligence Index | 30.4 | 29.2 | DeepSeek V4 Pro 0423 |
| Epoch Capabilities Index | 149.1 | 146.9 | DeepSeek V4 Pro 0423 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index35
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| AA Agentic Index | 26.3 | 29.5 | MiniMax M3 |
| Design Arena: agents building web apps | 1000 | 1200 | MiniMax M3 |
| Design Arena: agents building full-stack apps | 948 | 1197 | MiniMax M3 |
| Design Arena: agents building mobile apps | Not measured | 1169 | Only MiniMax M3 measured |
| Design Arena: agents building Android apps | Not measured | 1164 | Only MiniMax M3 measured |
| Design Arena: agents building Godot games | 1059 | Not measured | Only DeepSeek V4 Pro 0423 measured |
| Design Arena: agents building games | Not measured | 1158 | Only MiniMax M3 measured |
| Design Arena: HTML slides | Not measured | 1180 | Only MiniMax M3 measured |
| Design Arena: PowerPoint slides | Not measured | 1205 | Only MiniMax M3 measured |
| Design Arena: spreadsheets | Not measured | 1021 | Only MiniMax M3 measured |
| LiveBench research | Not measured | 68.0% | Only MiniMax M3 measured |
| LiveBench browser | Not measured | 8.0% | Only MiniMax M3 measured |
| LiveBench long agentic | Not measured | 12.4% | Only MiniMax M3 measured |
| Finance Agent (Vals AI) | 44.1% | 48.3% | MiniMax M3 |
| Tax Agent Bench (Vals AI) | 25.2% | 20.6% | DeepSeek V4 Pro 0423 |
| Legal Research Bench (Vals AI) | 23.1% | 29.8% | MiniMax M3 |
| Excel Modeling Benchmark (Vals AI) | 51.6% | 47.8% | DeepSeek V4 Pro 0423 |
| τ²-Bench Telecom | Not measured | 88.9% | Only MiniMax M3 measured |
| APEX-Agents | Not measured | 37.7% | Only MiniMax M3 measured |
| OSWorld 2.0 | Not measured | 4.6% | Only MiniMax M3 measured |
| Vending-Bench 2 | $3,285 | $2,158 | DeepSeek V4 Pro 0423 |
| Neither model has a result on the other 18 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index51
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| GPQA Diamond | 89.4% | 92.7% | MiniMax M3 |
| LiveBench reasoning | Not measured | 74.5% | Only MiniMax M3 measured |
| MMMU Pro (Vals AI) | Not measured | 81.2% | Only MiniMax M3 measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 92.9% | Only MiniMax M3 measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 39.0% | Only MiniMax M3 measured |
| GPQA Diamond (Epoch AI run) | 90.9% | 90.9% | Tie |
| SimpleBench | Not measured | 45.8% | Only MiniMax M3 measured |
| Chess puzzles | 19.0% | 13.0% | DeepSeek V4 Pro 0423 |
| Mystery game puzzles | 17.0% | 8.0% | DeepSeek V4 Pro 0423 |
| LMArena Text: hard prompts | 1461 | 1447 | DeepSeek V4 Pro 0423 |
| LMArena Text: expert | 1463 | 1461 | DeepSeek V4 Pro 0423 |
| LMArena Text: hard prompts in English | 1466 | 1455 | DeepSeek V4 Pro 0423 |
| LMArena Vision: diagrams | Not measured | 1266 | Only MiniMax M3 measured |
| LMArena Vision: homework | Not measured | 1268 | Only MiniMax M3 measured |
| Neither model has a result on the other 7 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index31
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| OTIS Mock AIME | 96.7% | 71.1% | DeepSeek V4 Pro 0423 |
| FrontierMath, tiers 1 to 3 | 45.3% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| FrontierMath, tier 4 | 2.4% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| ProofBench | 16.0% | 18.0% | MiniMax M3 |
| LMArena Text: math | 1443 | 1429 | DeepSeek V4 Pro 0423 |
| LMArena Text: mathematical work | 1455 | 1448 | DeepSeek V4 Pro 0423 |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index54
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| MMLU Pro | 87.3% | 84.2% | DeepSeek V4 Pro 0423 |
| LegalBench (Vals AI) | 80.3% | 85.4% | MiniMax M3 |
| CaseLaw (Vals AI) | 59.4% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| CorpFin (Vals AI) | 61.4% | 68.1% | MiniMax M3 |
| TaxEval (Vals AI) | 72.1% | 72.7% | MiniMax M3 |
| MortgageTax (Vals AI) | Not measured | 68.4% | Only MiniMax M3 measured |
| MedCode (Vals AI) | 40.5% | 46.3% | MiniMax M3 |
| SimpleQA Verified | 47.0% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| LMArena Text: science | 1470 | 1452 | DeepSeek V4 Pro 0423 |
| LMArena Text: medicine and healthcare | 1466 | 1441 | DeepSeek V4 Pro 0423 |
| LMArena Text: legal and government | 1464 | 1437 | DeepSeek V4 Pro 0423 |
| LMArena Text: business and finance | 1443 | 1434 | DeepSeek V4 Pro 0423 |
| Neither model has a result on the other 5 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index26
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| Design Arena: websites | 1235 | 1257 | MiniMax M3 |
| Design Arena: UI components | 1221 | 1246 | MiniMax M3 |
| Design Arena: game dev | 1229 | 1210 | DeepSeek V4 Pro 0423 |
| Design Arena: data viz | 1203 | 1235 | MiniMax M3 |
| Design Arena: 3D | 1254 | 1209 | DeepSeek V4 Pro 0423 |
| Design Arena: all code categories | 1240 | 1247 | MiniMax M3 |
| Design Arena: SVG | 1139 | 1161 | MiniMax M3 |
| Design Arena: ASCII art | 1148 | 1161 | MiniMax M3 |
| LiveBench design | Not measured | 66.2% | Only MiniMax M3 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index193
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | Leader |
|---|---|---|---|
| LiveBench data analysis | Not measured | 76.2% | Only MiniMax M3 measured |
| Public Benefits Bench (Vals AI) | 62.9% | 64.1% | MiniMax M3 |
| SAGE (Vals AI) | Not measured | 50.6% | Only MiniMax M3 measured |
| MedScribe (Vals AI) | 75.1% | 87.3% | MiniMax M3 |
| Vals Multimodal Index | Not measured | 60.0% | Only MiniMax M3 measured |
| IFBench | Not measured | 82.9% | Only MiniMax M3 measured |
| Long context reasoning (AA-LCR) | Not measured | 83.0% | Only MiniMax M3 measured |
| DTBench | 90.7% | 78.9% | DeepSeek V4 Pro 0423 |
| LMCA | 41.2% | 33.7% | DeepSeek V4 Pro 0423 |
| Surface Evolver Bench | 40.0% | 55.0% | MiniMax M3 |
| CL-bench Life | 13.5% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| LMArena Text | 1451 | 1433 | DeepSeek V4 Pro 0423 |
| LMArena Text: instruction following | 1445 | 1433 | DeepSeek V4 Pro 0423 |
| LMArena Text: creative writing | 1446 | 1404 | DeepSeek V4 Pro 0423 |
| LMArena Vision | Not measured | 1251 | Only MiniMax M3 measured |
| LMArena Text: English | 1459 | 1443 | DeepSeek V4 Pro 0423 |
| LMArena Text: non-English | 1439 | 1420 | DeepSeek V4 Pro 0423 |
| LMArena Text: Chinese | 1481 | 1466 | DeepSeek V4 Pro 0423 |
| LMArena Text: German | 1458 | 1428 | DeepSeek V4 Pro 0423 |
| LMArena Text: Japanese | 1446 | 1380 | DeepSeek V4 Pro 0423 |
| LMArena Text: Korean | 1409 | 1373 | DeepSeek V4 Pro 0423 |
| LMArena Text: French | 1473 | 1444 | DeepSeek V4 Pro 0423 |
| LMArena Text: Spanish | 1457 | 1433 | DeepSeek V4 Pro 0423 |
| LMArena Text: Russian | 1453 | 1426 | DeepSeek V4 Pro 0423 |
| LMArena Text: Polish | 1461 | 1423 | DeepSeek V4 Pro 0423 |
| LMArena Text: multi-turn | 1468 | 1443 | DeepSeek V4 Pro 0423 |
| LMArena Text: longer queries | 1458 | 1446 | DeepSeek V4 Pro 0423 |
| LMArena Text: writing and language | 1447 | 1419 | DeepSeek V4 Pro 0423 |
| LMArena Text: entertainment and media | 1431 | 1403 | DeepSeek V4 Pro 0423 |
| LMArena Vision: English | Not measured | 1263 | Only MiniMax M3 measured |
| LMArena Vision: Chinese | Not measured | 1311 | Only MiniMax M3 measured |
| LMArena Vision: reading text | Not measured | 1259 | Only MiniMax M3 measured |
| LMArena Vision: humor | Not measured | 1235 | Only MiniMax M3 measured |
| LMArena Vision: recognizing things | Not measured | 1197 | Only MiniMax M3 measured |
| LMArena Vision: creative writing | Not measured | 1234 | Only MiniMax M3 measured |
| LMArena Document | Not measured | 1435 | Only MiniMax M3 measured |
| Neither model has a result on the other 10 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | DeepSeek V4 Pro 0423 | MiniMax M3 | 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
MiniMax M3 costs less
Repository review
50K input + 3K output tokens
MiniMax M3 costs less
Agent loop
200K input + 10K output tokens
MiniMax M3 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 Pro 0423 | MiniMax M3 |
|---|---|---|
| Maker | DeepSeek | MiniMax |
| Weights | Open weights | Open weights |
| Released | April 24, 2026 | May 31, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $0.9553 | $0.30 |
| Output $/Mtok | $1.9105 | $1.20 |
| Coding index | 35.3 | 57.8 |
| Independent measurements behind it | 1 | 2 |
| Cost per solved task | $0.57 | $0.56 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is DeepSeek V4 Pro 0423 better than MiniMax M3 for coding?
Pick MiniMax M3 for coding: it scores 57.8 on the Vibe Coding Index against 35.3 for DeepSeek V4 Pro 0423. Pick DeepSeek V4 Pro 0423 when what matters is AA Coding Index or LiveCodeBench (Vals AI), where it leads.
Which is cheaper, DeepSeek V4 Pro 0423 or MiniMax M3?
DeepSeek V4 Pro 0423 costs $0.9553 per million input tokens and $1.9105 per million output tokens; MiniMax M3 costs $0.30 and $1.20. Per solved SWE-bench Verified task, DeepSeek V4 Pro 0423 costs $0.57 and MiniMax M3 $0.56.
Which has the bigger context window, DeepSeek V4 Pro 0423 or MiniMax M3?
Neither: both take 1.05M tokens.
How many benchmarks is the DeepSeek V4 Pro 0423 vs MiniMax M3 comparison based on?
75 benchmarks have a result for both models as of October 21, 2026. DeepSeek V4 Pro 0423's coding index rests on 1 independent benchmark and MiniMax M3's on 2. Where only one model has a result, the page says so instead of naming a winner.