Model comparison
DeepSeek-V3 vs o4-mini
o4-mini is the stronger model overall, scoring 41.6 to 39.5 on the Noometry Index. DeepSeek-V3 costs 4.8× less per token, which makes it the better buy when o4-mini's lead doesn't matter for your workload.
Last verified . 40 shared benchmarks.
DeepSeekSummary
- They share 40 benchmarks with published results for both. DeepSeek-V3 scores higher in 3 categories and o4-mini in 5 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in long context, where o4-mini leads 45.5 to 34.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 81.7% for o4-mini.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $1.10 / $4.40 for o4-mini.
- o4-mini accepts more context: 200K tokens versus 164K.
- DeepSeek-V3 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3 | o4-mini | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 39.5 | 41.6 |
| Released | 2024-12-26 | 2025-04-16 |
| Weights | Open | Proprietary |
| Context window | 164K | 200K |
| Max output | 164K | 100K |
| Input $ / M tokens | $0.24 | $1.10 |
| Output $ / M tokens | $0.90 | $4.40 |
| Results tracked | 60 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), o4-mini: 40.9 (#127)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| Aider Polyglot | 55.1% | 72% |
| WeirdML | 36.1% | 52.6% |
| LMArena Coding | 1368 | 1368 |
| SWE-bench Verified (bash only) | — | 45% |
| SciCode | 35.8% | — |
| GSO | — | 3.6% |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| CadEval | — | 62% |
| ALE-Bench | — | 826.17 |
| AlgoTune | — | 1.72 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, o4-mini: 32.6 (#61)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| METR Time Horizons | 49.6% | 63.9% |
| Berkeley Function Calling Leaderboard | — | 53.2% |
| GDPval | — | 25.3% |
Reasoning o4-mini leads
DeepSeek-V3: 20.5 (#236), o4-mini: 24.6 (#162)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| SimpleBench | 27.2% | 38.7% |
| Kagi LLM Benchmark | 52.3% | 67.6% |
| CritPt | 0% | 0.6% |
| LMArena Hard Prompts | 1365 | 1351 |
| DTBench | 64.8% | 77.6% |
| LMCA | 15.5% | 26.5% |
| Epoch Capabilities Index | 135.94 | 145.64 |
| ForecastBench | 59.1 | 61.8 |
| ARC-AGI-2 | — | 6.1% |
| ARC-AGI-1 | — | 58.7% |
| Chess Puzzles | — | 26% |
| EnigmaEval | — | 9.2% |
| LiveBench Reasoning | 65.8% | — |
| Mystery Game Puzzles | — | 5% |
| LiveBench Data Analysis | 60.9% | — |
| BIG-Bench Hard | 87.5% | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math o4-mini leads
DeepSeek-V3: 32.1 (#219), o4-mini: 40.8 (#89)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 81.7% |
| Omni-MATH | 40.3% | 72% |
| LMArena Math | 1373 | 1389 |
| MATH Level 5 | 75.5% | 97.8% |
| FrontierMath (Feb 2025 set) | 1.7% | 24.8% |
| FrontierMath (Tiers 1-3) | — | 36.1% |
| FrontierMath Tier 4 | — | 4.9% |
| LiveBench Math | 73.5% | — |
| FrontierMath Tier 4 (v1) | — | 6.3% |
Knowledge o4-mini leads
DeepSeek-V3: 37.5 (#155), o4-mini: 43.6 (#91)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| GPQA Diamond | 67.6% | 79.6% |
| MMLU-Pro | 72.3% | 82% |
| Confabulations | 26.1% | 15.8% |
| Vectara Hallucination Rate | 6.1% | 18.6% |
| GPQA (HELM) | 53.8% | 73.5% |
| LMArena Expert | 1351 | 1343 |
| Humanity's Last Exam | — | 18.1% |
| SimpleQA Verified | — | 19.6% |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multimodal Not comparable
DeepSeek-V3: —, o4-mini: 40.2 (#49)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| LMArena Vision | — | 1194 |
| GeoBench | — | 64% |
| VPCT | — | 57.5% |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), o4-mini: 47.0 (#154)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| LMArena Non-English | 1358 | 1337 |
| LMArena Chinese | 1391 | 1354 |
| LMArena French | 1385 | 1364 |
| LMArena German | 1374 | 1336 |
| LMArena Japanese | 1333 | 1308 |
| LMArena Korean | 1319 | 1312 |
| LMArena Russian | 1373 | 1334 |
| LMArena Spanish | 1358 | 1347 |
Instruction Following o4-mini leads
DeepSeek-V3: 72.8 (#130), o4-mini: 75.2 (#68)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| IFEval | 83.2% | 92.8% |
| LMArena Instruction Following | 1345 | 1321 |
| LiveBench Instruction Following | 81.5% | — |
Long Context o4-mini leads
DeepSeek-V3: 34.0 (#253), o4-mini: 45.5 (#33)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| Fiction.LiveBench | 50% | 77.8% |
| LMArena Longer Query | 1352 | 1315 |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), o4-mini: 54.0 (#152)
| Benchmark | DeepSeek-V3 | o4-mini |
|---|---|---|
| LMArena Text | 1375 | 1353 |
| LMArena Creative Writing | 1364 | 1294 |
| Short-Story Creative Writing | 77% | 75% |
| WildBench | 83% | 85.4% |
| LMArena Multi-Turn | 1389 | 1350 |
| EQ-Bench Creative Writing | 1472 | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than o4-mini?
o4-mini is the stronger model overall, scoring 41.6 to 39.5 on the Noometry Index. DeepSeek-V3 costs 4.8× less per token, which makes it the better buy when o4-mini's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3 or o4-mini?
DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; o4-mini lists at $1.10 and $4.40.
Is DeepSeek-V3 or o4-mini better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 40.9 in the Noometry coding category.
Which has the bigger context window?
o4-mini does, with 200K tokens against 164K.
How many benchmarks do DeepSeek-V3 and o4-mini share?
40 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and o4-mini has 60.