Model comparison
DeepSeek V4 Pro vs o3-mini
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 36.7 on the Noometry Index.
Last verified . 33 shared benchmarks.
DeepSeekSummary
- They share 33 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and o3-mini in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.5 to 16.3.
- The biggest single-benchmark swing is ARC-AGI-2: 61.3% for DeepSeek V4 Pro and 3% for o3-mini.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $1.10 / $4.40 for o3-mini.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 200K.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Pro | o3-mini | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 54.3 | 36.7 |
| Released | 2026-04-24 | 2024-12-20 |
| Weights | Open | Proprietary |
| Context window | 1M | 200K |
| Max output | 393K | 100K |
| Input $ / M tokens | $0.66 | $1.10 |
| Output $ / M tokens | $1.98 | $4.40 |
| Results tracked | 48 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
DeepSeek V4 Pro: 52.4 (#34), o3-mini: 40.8 (#132)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| SciCode | 51% | 39.8% |
| WeirdML | 66.2% | 43.7% |
| LMArena Coding | 1470 | 1378 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| Aider Polyglot | — | 60.4% |
| LMArena WebDev | 1582 | — |
| GSO | — | 1.3% |
| LiveBench Coding | — | 82.7% |
| CadEval | — | 54% |
| ALE-Bench | 1,403 | — |
Agentic & Tool Use DeepSeek V4 Pro leads
DeepSeek V4 Pro: 32.8 (#58), o3-mini: 29.6 (#84)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| APEX-Agents | 47.3% | — |
| Cybench | — | 22.5% |
| Vending-Bench 2 | 3,285 | — |
Reasoning DeepSeek V4 Pro leads
DeepSeek V4 Pro: 56.5 (#24), o3-mini: 16.3 (#305)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| ARC-AGI-2 | 61.3% | 3% |
| ARC-AGI-1 | 90.5% | 34.5% |
| CritPt | 18% | 0.3% |
| Chess Puzzles | 47% | 17% |
| LMArena Hard Prompts | 1461 | 1366 |
| Mystery Game Puzzles | 43% | 7% |
| DTBench | 93.9% | 68.8% |
| LMCA | 45.5% | 19% |
| Epoch Capabilities Index | 155.31 | 140.34 |
| ForecastBench | 56.1 | 59.6 |
| SimpleBench | — | 22.8% |
| Kagi LLM Benchmark | 53.5% | — |
| NYT Connections (extended) | 91.3% | — |
| LiveBench Reasoning | — | 89.6% |
| LiveBench Data Analysis | — | 70.6% |
| Surface Evolver Bench | 40% | — |
| LiveBench | — | 75.9% |
Math DeepSeek V4 Pro leads
DeepSeek V4 Pro: 64.8 (#30), o3-mini: 28.1 (#244)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 18.6% |
| FrontierMath Tier 4 | 26.8% | 0% |
| OTIS Mock AIME 2024-2025 | 98.6% | 76.9% |
| LMArena Math | 1455 | 1396 |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| LiveBench Math | — | 77.3% |
| MATH Level 5 | — | 96.5% |
| FrontierMath (Feb 2025 set) | — | 12.4% |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge DeepSeek V4 Pro leads
DeepSeek V4 Pro: 59.5 (#31), o3-mini: 38.3 (#146)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| GPQA Diamond | 91.7% | 77% |
| SimpleQA Verified | 52.9% | 15.3% |
| LMArena Expert | 1464 | 1364 |
| Confabulations | — | 17.9% |
| Vectara Hallucination Rate | 8.6% | — |
Multilingual DeepSeek V4 Pro leads
DeepSeek V4 Pro: 54.4 (#45), o3-mini: 45.7 (#164)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| LMArena Non-English | 1439 | 1319 |
| LMArena Chinese | 1486 | 1379 |
| LMArena French | 1472 | 1334 |
| LMArena German | 1458 | 1303 |
| LMArena Japanese | 1445 | 1286 |
| LMArena Korean | 1447 | 1314 |
| LMArena Russian | 1453 | 1304 |
| LMArena Spanish | 1458 | 1321 |
Instruction Following DeepSeek V4 Pro leads
DeepSeek V4 Pro: 76.1 (#47), o3-mini: 75.1 (#72)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| LMArena Instruction Following | 1448 | 1337 |
| LiveBench Instruction Following | — | 84.4% |
Long Context DeepSeek V4 Pro leads
DeepSeek V4 Pro: 45.0 (#51), o3-mini: 33.8 (#256)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| LMArena Longer Query | 1458 | 1343 |
| Fiction.LiveBench | — | 50% |
| CL-bench Life | 13.5% | — |
Writing & Preference DeepSeek V4 Pro leads
DeepSeek V4 Pro: 65.5 (#46), o3-mini: 50.3 (#182)
| Benchmark | DeepSeek V4 Pro | o3-mini |
|---|---|---|
| LMArena Text | 1451 | 1337 |
| LMArena Creative Writing | 1446 | 1286 |
| LMArena Multi-Turn | 1467 | 1320 |
| Short-Story Creative Writing | — | 61.7% |
| EQ-Bench Creative Writing | 1553 | — |
| EQ-Bench 4 | 1166 | — |
| LiveBench Language | — | 50.7% |
Frequently asked questions
Is DeepSeek V4 Pro better than o3-mini?
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 36.7 on the Noometry Index.
Which is cheaper, DeepSeek V4 Pro or o3-mini?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; o3-mini lists at $1.10 and $4.40.
Is DeepSeek V4 Pro or o3-mini better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 40.8 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Pro does, with 1M tokens against 200K.
How many benchmarks do DeepSeek V4 Pro and o3-mini share?
33 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and o3-mini has 51.