Model comparison
DeepSeek-R1 vs o3-mini
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 36.7 on the Noometry Index.
Last verified . 39 shared benchmarks.
DeepSeekSummary
- They share 39 benchmarks with published results for both. DeepSeek-R1 scores higher in 8 categories and o3-mini in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-R1 leads 43.8 to 28.1.
- The biggest single-benchmark swing is Fiction.LiveBench: 75% for DeepSeek-R1 and 50% for o3-mini.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $1.10 / $4.40 for o3-mini.
- o3-mini accepts more context: 200K tokens versus 164K.
Side by side
| DeepSeek-R1 | o3-mini | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.3 | 36.7 |
| Released | 2025-01-20 | 2024-12-20 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | 200K |
| Max output | 64K | 100K |
| Input $ / M tokens | $0.50 | $1.10 |
| Output $ / M tokens | $2.15 | $4.40 |
| Results tracked | 52 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), o3-mini: 40.8 (#132)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| Aider Polyglot | 71.4% | 60.4% |
| SciCode | 35.7% | 39.8% |
| WeirdML | 41.6% | 43.7% |
| LiveBench Coding | 66.7% | 82.7% |
| LMArena Coding | 1427 | 1378 |
| GSO | — | 1.3% |
| CadEval | — | 54% |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), o3-mini: 29.6 (#84)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| Cybench | — | 22.5% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning DeepSeek-R1 leads
DeepSeek-R1: 18.6 (#278), o3-mini: 16.3 (#305)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| ARC-AGI-2 | 1.3% | 3% |
| SimpleBench | 40.8% | 22.8% |
| ARC-AGI-1 | 21.2% | 34.5% |
| CritPt | 1.1% | 0.3% |
| LiveBench Reasoning | 83.2% | 89.6% |
| LMArena Hard Prompts | 1416 | 1366 |
| LiveBench Data Analysis | 69.8% | 70.6% |
| Epoch Capabilities Index | 141.29 | 140.34 |
| ForecastBench | 60 | 59.6 |
| LiveBench | 71.6% | 75.9% |
| Kagi LLM Benchmark | 69.4% | — |
| Chess Puzzles | — | 17% |
| Mystery Game Puzzles | — | 7% |
| DTBench | — | 68.8% |
| LMCA | — | 19% |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), o3-mini: 28.1 (#244)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 76.9% |
| LiveBench Math | 80.7% | 77.3% |
| LMArena Math | 1400 | 1396 |
| MATH Level 5 | 96.6% | 96.5% |
| FrontierMath (Tiers 1-3) | — | 18.6% |
| FrontierMath Tier 4 | — | 0% |
| Omni-MATH | 42.4% | — |
| FrontierMath (Feb 2025 set) | — | 12.4% |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), o3-mini: 38.3 (#146)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| GPQA Diamond | 76.3% | 77% |
| Confabulations | 12.7% | 17.9% |
| LMArena Expert | 1394 | 1364 |
| SimpleQA Verified | — | 15.3% |
| MMLU-Pro | 79.3% | — |
| Vectara Hallucination Rate | 11.3% | — |
| GPQA (HELM) | 66.6% | — |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), o3-mini: 45.7 (#164)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| LMArena Non-English | 1412 | 1319 |
| LMArena Chinese | 1442 | 1379 |
| LMArena French | 1417 | 1334 |
| LMArena German | 1404 | 1303 |
| LMArena Japanese | 1391 | 1286 |
| LMArena Korean | 1360 | 1314 |
| LMArena Russian | 1423 | 1304 |
| LMArena Spanish | 1411 | 1321 |
Instruction Following o3-mini leads
DeepSeek-R1: 72.0 (#143), o3-mini: 75.1 (#72)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| LiveBench Instruction Following | 80.5% | 84.4% |
| LMArena Instruction Following | 1382 | 1337 |
| IFEval | 78.4% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), o3-mini: 33.8 (#256)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| Fiction.LiveBench | 75% | 50% |
| LMArena Longer Query | 1391 | 1343 |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), o3-mini: 50.3 (#182)
| Benchmark | DeepSeek-R1 | o3-mini |
|---|---|---|
| LMArena Text | 1428 | 1337 |
| LMArena Creative Writing | 1405 | 1286 |
| Short-Story Creative Writing | 83% | 61.7% |
| LMArena Multi-Turn | 1405 | 1320 |
| LiveBench Language | 48.5% | 50.7% |
| EQ-Bench Creative Writing | 1500 | — |
| WildBench | 82.8% | — |
Frequently asked questions
Is DeepSeek-R1 better than o3-mini?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 36.7 on the Noometry Index.
Which is cheaper, DeepSeek-R1 or o3-mini?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; o3-mini lists at $1.10 and $4.40.
Is DeepSeek-R1 or o3-mini better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 40.8 in the Noometry coding category.
Which has the bigger context window?
o3-mini does, with 200K tokens against 164K.
How many benchmarks do DeepSeek-R1 and o3-mini share?
39 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and o3-mini has 51.