DeepSeek V4 Pro 0423 vs GPT-5.6 Luna
GPT-5.6 Luna leads DeepSeek V4 Pro 0423 on coding, 58.0 to 38.7 on the Vibe Coding Index, and GPT-5.6 Luna costs 2.5x less per output token ($1.20 vs $2.976 per million). Updated October 22, 2026; 78 benchmarks have a result for both.
DeepSeek V4 Pro 0423 or GPT-5.6 Luna: which to pick
Pick GPT-5.6 Luna for coding: it scores 58.0 on the Vibe Coding Index against 38.7 for DeepSeek V4 Pro 0423.
Pick DeepSeek V4 Pro 0423 for work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.
- Shared results
- 78
- DeepSeek V4 Pro 0423 only
- 6
- GPT-5.6 Luna only
- 41
- 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
GPT-5.6 Luna
GPT-5.6 Luna scores 58.0 against 38.7.
5 and 6 benchmarks behind the scores
Intelligence
AA Intelligence Index
GPT-5.6 Luna
GPT-5.6 Luna leads, 37.3 against 30.4.
Same board, both measured
Agentic work
AA Agentic Index
GPT-5.6 Luna
GPT-5.6 Luna leads, 42.1 against 26.3.
Same board, both measured
Reasoning
GPQA Diamond
GPT-5.6 Luna
GPT-5.6 Luna leads, 91.7% against 89.4%.
Same board, both measured
Knowledge
MMLU Pro
DeepSeek V4 Pro 0423
DeepSeek V4 Pro 0423 leads, 87.3% against 86.0%.
Same board, both measured
Web design
Design Arena: websites
GPT-5.6 Luna
GPT-5.6 Luna leads, 1255 against 1234.
Same board, both measured
Long documents
Listed context window
GPT-5.6 Luna
GPT-5.6 Luna takes 1.05M tokens against 1.05M.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
GPT-5.6 Luna
DeepSeek V4 Pro 0423 $0.0028, GPT-5.6 Luna $0.0008 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
GPT-5.6 Luna
DeepSeek V4 Pro 0423 $0.57, GPT-5.6 Luna $0.04.
Published attempt cost and pass rate
At a glance
- Coding index
- DeepSeek V4 Pro 0423 38.7 from 5 benchmarks, GPT-5.6 Luna 58.0 from 6 benchmarks.
- Widest category
- Math, 24.4 points apart on average over 6 shared benchmarks.
- Widest single gap
- FrontierMath, tier 4: DeepSeek V4 Pro 0423 2.4% against GPT-5.6 Luna 61.0%, 58.5 points apart on a 0 to 100 scale.
- Output price
- DeepSeek V4 Pro 0423 $2.976 against GPT-5.6 Luna $1.20 per million output tokens, 2.5x apart.
- Context window
- DeepSeek V4 Pro 0423 takes 1.05M tokens, GPT-5.6 Luna 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 V4 Pro 0423 | Bars | GPT-5.6 Luna | Ties |
|---|---|---|---|---|
| Coding | 2 | 13 | 1 tie | |
| Intelligence | 0 | 2 | ||
| Agentic work | 1 | 5 | ||
| Reasoning | 3 | 5 | ||
| Math | 0 | 6 | ||
| Knowledge | 6 | 4 | ||
| Web design | 4 | 4 | ||
| General | 18 | 4 | ||
| All shared boards | 34 | 43 | 1 tie |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index213
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| LiveBench coding | Not measured | 73.6% | Only GPT-5.6 Luna measured |
| LMArena WebDev | 1447 | 1518 | GPT-5.6 Luna |
| AA Coding Index | 59.4 | 71.4 | GPT-5.6 Luna |
| LiveCodeBench (Vals AI) | 87.5% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| SWE-bench Verified (Vals AI) | 77.4% | 93.0% | GPT-5.6 Luna |
| Terminal-Bench 4 (Vals AI) | 11.1% | 11.6% | GPT-5.6 Luna |
| Vibe Code Bench (Vals AI) | 49.9% | 77.1% | GPT-5.6 Luna |
| Code migration (Vals AI) | 26.2% | 44.5% | GPT-5.6 Luna |
| ProgramBench (Vals AI) | 0.0% | 0.0% | Tie |
| IOI (Vals AI) | Not measured | 61.8% | Only GPT-5.6 Luna measured |
| Terminal-Bench 2.1 (Vals AI) | 50.2% | 79.0% | GPT-5.6 Luna |
| Vibe Code Bench 1-100 (Vals AI) | Not measured | 22.6% | Only GPT-5.6 Luna measured |
| SkillsBench (Vals AI) | 51.3% | 60.5% | GPT-5.6 Luna |
| CyberBench (Vals AI) | Not measured | 73.6% | Only GPT-5.6 Luna measured |
| SciCode | Not measured | 53.6% | Only GPT-5.6 Luna measured |
| FrontierCode | 17.6% | 39.8% | GPT-5.6 Luna |
| WeirdML | 48.9% | 60.9% | GPT-5.6 Luna |
| ALE-Bench | 1006.1 | 1667.4 | GPT-5.6 Luna |
| LMArena Text: coding | 1470 | 1466 | DeepSeek V4 Pro 0423 |
| LMArena Text: software and IT | 1469 | 1463 | DeepSeek V4 Pro 0423 |
| LMArena WebDev: React | 1441 | 1508 | GPT-5.6 Luna |
| LMArena WebDev: HTML | 1435 | 1520 | GPT-5.6 Luna |
| LMArena WebDev: from an image | Not measured | 1517 | Only GPT-5.6 Luna measured |
| SciCode | Not measured | 6.2% | Only GPT-5.6 Luna measured |
| Neither model has a result on the other 15 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| AA Intelligence Index | 30.4 | 37.3 | GPT-5.6 Luna |
| Epoch Capabilities Index | 149.1 | 156.5 | GPT-5.6 Luna |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index15
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| AA Agentic Index | 26.3 | 42.1 | GPT-5.6 Luna |
| Design Arena: agents building web apps | 1000 | Not measured | Only DeepSeek V4 Pro 0423 measured |
| Design Arena: agents building full-stack apps | 948 | Not measured | Only DeepSeek V4 Pro 0423 measured |
| Design Arena: agents building Godot games | 1059 | Not measured | Only DeepSeek V4 Pro 0423 measured |
| LiveBench research | Not measured | 51.0% | Only GPT-5.6 Luna measured |
| LiveBench browser | Not measured | 68.0% | Only GPT-5.6 Luna measured |
| LiveBench long agentic | Not measured | 30.4% | Only GPT-5.6 Luna measured |
| SREBench (Vals AI) | Not measured | 1.5% | Only GPT-5.6 Luna measured |
| Finance Agent (Vals AI) | 44.1% | 55.0% | GPT-5.6 Luna |
| Tax Agent Bench (Vals AI) | 25.2% | 24.8% | DeepSeek V4 Pro 0423 |
| Legal Research Bench (Vals AI) | 23.1% | 36.5% | GPT-5.6 Luna |
| Excel Modeling Benchmark (Vals AI) | 51.6% | 67.1% | GPT-5.6 Luna |
| BioMysteryBench (Vals AI) | Not measured | 61.5% | Only GPT-5.6 Luna measured |
| Terminal-Bench-Science (Vals AI) | Not measured | 0.0% | Only GPT-5.6 Luna measured |
| Vals Science Index | Not measured | 25.3% | Only GPT-5.6 Luna measured |
| APEX-Agents | Not measured | 43.0% | Only GPT-5.6 Luna measured |
| BALROG | Not measured | 45.6% | Only GPT-5.6 Luna measured |
| Vending-Bench 2 | $3,285 | $4,095 | GPT-5.6 Luna |
| Neither model has a result on the other 23 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index35
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| GPQA Diamond | 89.4% | 91.7% | GPT-5.6 Luna |
| LiveBench reasoning | Not measured | 85.6% | Only GPT-5.6 Luna measured |
| MysteryMechanism (Vals AI) | Not measured | 14.4% | Only GPT-5.6 Luna measured |
| MMMU Pro (Vals AI) | Not measured | 85.0% | Only GPT-5.6 Luna measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 91.1% | Only GPT-5.6 Luna measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 39.5% | Only GPT-5.6 Luna measured |
| GPQA Diamond (Epoch AI run) | 90.9% | 91.6% | GPT-5.6 Luna |
| ARC-AGI | Not measured | 88.0% | Only GPT-5.6 Luna measured |
| ARC-AGI-2 | Not measured | 59.5% | Only GPT-5.6 Luna measured |
| SimpleBench | Not measured | 46.8% | Only GPT-5.6 Luna measured |
| Chess puzzles | 19.0% | 43.0% | GPT-5.6 Luna |
| Mystery game puzzles | 17.0% | 21.0% | GPT-5.6 Luna |
| LMArena Text: hard prompts | 1461 | 1451 | DeepSeek V4 Pro 0423 |
| LMArena Text: expert | 1463 | 1478 | GPT-5.6 Luna |
| LMArena Text: hard prompts in English | 1466 | 1460 | DeepSeek V4 Pro 0423 |
| LMArena Vision: diagrams | Not measured | 1273 | Only GPT-5.6 Luna measured |
| LMArena Vision: homework | Not measured | 1282 | Only GPT-5.6 Luna measured |
| Kagi LLM Benchmark | 53.5% | 49.1% | DeepSeek V4 Pro 0423 |
| Neither model has a result on the other 4 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| OTIS Mock AIME | 96.7% | 98.3% | GPT-5.6 Luna |
| FrontierMath, tiers 1 to 3 | 45.3% | 82.1% | GPT-5.6 Luna |
| FrontierMath, tier 4 | 2.4% | 61.0% | GPT-5.6 Luna |
| ProofBench | 16.0% | 60.0% | GPT-5.6 Luna |
| LMArena Text: math | 1443 | 1459 | GPT-5.6 Luna |
| LMArena Text: mathematical work | 1455 | 1470 | GPT-5.6 Luna |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index64
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| MMLU Pro | 87.3% | 86.0% | DeepSeek V4 Pro 0423 |
| LegalBench (Vals AI) | 80.3% | 84.0% | GPT-5.6 Luna |
| CaseLaw (Vals AI) | 59.4% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| CorpFin (Vals AI) | 61.4% | 64.2% | GPT-5.6 Luna |
| TaxEval (Vals AI) | 72.1% | 76.2% | GPT-5.6 Luna |
| MortgageTax (Vals AI) | Not measured | 67.3% | Only GPT-5.6 Luna measured |
| MedCode (Vals AI) | 40.5% | 42.4% | GPT-5.6 Luna |
| SimpleQA Verified | 47.0% | 41.0% | DeepSeek V4 Pro 0423 |
| LMArena Text: science | 1470 | 1433 | DeepSeek V4 Pro 0423 |
| LMArena Text: medicine and healthcare | 1466 | 1399 | DeepSeek V4 Pro 0423 |
| LMArena Text: legal and government | 1464 | 1446 | DeepSeek V4 Pro 0423 |
| LMArena Text: business and finance | 1443 | 1439 | DeepSeek V4 Pro 0423 |
| FACTS Benchmark Suite | Not measured | 35.0% | Only GPT-5.6 Luna measured |
| Neither model has a result on the other 5 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index44
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| Design Arena: websites | 1234 | 1255 | GPT-5.6 Luna |
| Design Arena: UI components | 1219 | 1248 | GPT-5.6 Luna |
| Design Arena: game dev | 1228 | 1194 | DeepSeek V4 Pro 0423 |
| Design Arena: data viz | 1203 | 1211 | GPT-5.6 Luna |
| Design Arena: 3D | 1253 | 1121 | DeepSeek V4 Pro 0423 |
| Design Arena: all code categories | 1240 | 1231 | DeepSeek V4 Pro 0423 |
| Design Arena: SVG | 1139 | 1194 | GPT-5.6 Luna |
| Design Arena: ASCII art | 1149 | 1142 | DeepSeek V4 Pro 0423 |
| LiveBench design | Not measured | 76.0% | Only GPT-5.6 Luna measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index184
| Benchmark | DeepSeek V4 Pro 0423 | GPT-5.6 Luna | Leader |
|---|---|---|---|
| LiveBench data analysis | Not measured | 78.0% | Only GPT-5.6 Luna measured |
| Public Benefits Bench (Vals AI) | 62.9% | 61.2% | DeepSeek V4 Pro 0423 |
| SAGE (Vals AI) | Not measured | 44.2% | Only GPT-5.6 Luna measured |
| MedScribe (Vals AI) | 75.1% | 84.4% | GPT-5.6 Luna |
| Vals Multimodal Index | Not measured | 69.1% | Only GPT-5.6 Luna measured |
| Long context reasoning (AA-LCR) | Not measured | 83.7% | Only GPT-5.6 Luna measured |
| DTBench | 90.7% | 89.1% | DeepSeek V4 Pro 0423 |
| LMCA | 41.2% | 48.5% | GPT-5.6 Luna |
| Surface Evolver Bench | 40.0% | 61.9% | GPT-5.6 Luna |
| Furniture assembly | Not measured | 42.5% | Only GPT-5.6 Luna measured |
| CL-bench Life | 13.5% | Not measured | Only DeepSeek V4 Pro 0423 measured |
| LMArena Text | 1451 | 1431 | DeepSeek V4 Pro 0423 |
| LMArena Text: instruction following | 1445 | 1437 | DeepSeek V4 Pro 0423 |
| LMArena Text: creative writing | 1446 | 1396 | DeepSeek V4 Pro 0423 |
| LMArena Vision | Not measured | 1257 | Only GPT-5.6 Luna measured |
| LMArena Text: English | 1459 | 1441 | DeepSeek V4 Pro 0423 |
| LMArena Text: non-English | 1439 | 1417 | DeepSeek V4 Pro 0423 |
| LMArena Text: Chinese | 1481 | 1470 | DeepSeek V4 Pro 0423 |
| LMArena Text: German | 1458 | 1455 | DeepSeek V4 Pro 0423 |
| LMArena Text: Japanese | 1446 | 1415 | DeepSeek V4 Pro 0423 |
| LMArena Text: Korean | 1409 | 1415 | GPT-5.6 Luna |
| LMArena Text: French | 1473 | 1458 | DeepSeek V4 Pro 0423 |
| LMArena Text: Spanish | 1457 | 1446 | DeepSeek V4 Pro 0423 |
| LMArena Text: Russian | 1453 | 1427 | DeepSeek V4 Pro 0423 |
| LMArena Text: Polish | 1461 | 1405 | DeepSeek V4 Pro 0423 |
| LMArena Text: multi-turn | 1468 | 1434 | DeepSeek V4 Pro 0423 |
| LMArena Text: longer queries | 1458 | 1436 | DeepSeek V4 Pro 0423 |
| LMArena Text: writing and language | 1447 | 1419 | DeepSeek V4 Pro 0423 |
| LMArena Text: entertainment and media | 1431 | 1402 | DeepSeek V4 Pro 0423 |
| LMArena Vision: English | Not measured | 1263 | Only GPT-5.6 Luna measured |
| LMArena Vision: Chinese | Not measured | 1294 | Only GPT-5.6 Luna measured |
| LMArena Vision: reading text | Not measured | 1264 | Only GPT-5.6 Luna measured |
| LMArena Vision: humor | Not measured | 1190 | Only GPT-5.6 Luna measured |
| LMArena Vision: creative writing | Not measured | 1257 | Only GPT-5.6 Luna measured |
| LMArena Document | Not measured | 1457 | Only GPT-5.6 Luna 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 V4 Pro 0423 | GPT-5.6 Luna | 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
GPT-5.6 Luna costs less
Repository review
50K input + 3K output tokens
GPT-5.6 Luna costs less
Agent loop
200K input + 10K output tokens
GPT-5.6 Luna 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 | OpenAI GPT-5.6 Luna |
|---|---|---|
| Maker | DeepSeek | OpenAI |
| Weights | Open weights | No public weights listed |
| Released | April 24, 2026 | July 9, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $1.358 | $0.20 |
| Output $/Mtok | $2.976 | $1.20 |
| Coding index | 38.7 | 58.0 |
| Independent measurements behind it | 5 | 6 |
| Cost per solved task | $0.57 | $0.04 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is DeepSeek V4 Pro 0423 better than GPT-5.6 Luna for coding?
Pick GPT-5.6 Luna for coding: it scores 58.0 on the Vibe Coding Index against 38.7 for DeepSeek V4 Pro 0423. Pick DeepSeek V4 Pro 0423 for work measured by LMArena Text: coding or LMArena Text: software and IT, where it leads.
Which is cheaper, DeepSeek V4 Pro 0423 or GPT-5.6 Luna?
DeepSeek V4 Pro 0423 costs $1.358 per million input tokens and $2.976 per million output tokens; GPT-5.6 Luna costs $0.20 and $1.20. Per solved SWE-bench Verified task, DeepSeek V4 Pro 0423 costs $0.57 and GPT-5.6 Luna $0.04.
Which has the bigger context window, DeepSeek V4 Pro 0423 or GPT-5.6 Luna?
GPT-5.6 Luna, with 1.05M tokens against 1.05M.
How many benchmarks is the DeepSeek V4 Pro 0423 vs GPT-5.6 Luna comparison based on?
78 benchmarks have a result for both models as of October 22, 2026. DeepSeek V4 Pro 0423's coding index rests on 5 independent benchmarks and GPT-5.6 Luna's on 6. Where only one model has a result, the page says so instead of naming a winner.