Model comparison
DeepSeek-V2.5 (Sep 2024) vs o1-mini
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 34.0 on the Noometry Index.
Last verified . 20 shared benchmarks.
DeepSeekSummary
- They share 20 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 4 categories and o1-mini in 4 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek-V2.5 (Sep 2024) leads 25.6 to 8.8.
- The biggest single-benchmark swing is Aider Polyglot: 17.8% for DeepSeek-V2.5 (Sep 2024) and 32.9% for o1-mini.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V2.5 (Sep 2024) | o1-mini | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 37.6 | 34.0 |
| Released | 2024-09-06 | 2024-09-12 |
| Weights | Open | Proprietary |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 22 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1-mini leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), o1-mini: 35.5 (#224)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| Aider Polyglot | 17.8% | 32.9% |
| LMArena Coding | 1309 | 1362 |
| HumanEval+ | 83.5% | 89% |
| MBPP+ | 74.1% | 78.8% |
| WeirdML | — | 36.3% |
| BigCodeBench Instruct | 48.6% | — |
| LiveBench Coding | — | 48% |
| BigCodeBench Complete | 53.2% | — |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, o1-mini: 24.6 (#118)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| Cybench | — | 10% |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), o1-mini: 8.8 (#346)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1333 |
| ARC-AGI-2 | — | 0.8% |
| SimpleBench | — | 18.1% |
| ARC-AGI-1 | — | 14% |
| LiveBench Reasoning | — | 72.3% |
| LiveBench Data Analysis | — | 57.9% |
| Epoch Capabilities Index | — | 135.82 |
| LiveBench | — | 57.8% |
Math Too close to call
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), o1-mini: 35.4 (#186)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| LMArena Math | 1288 | 1358 |
| OTIS Mock AIME 2024-2025 | — | 46.9% |
| LiveBench Math | — | 62% |
| MATH Level 5 | — | 89.2% |
| FrontierMath (Feb 2025 set) | — | 1.7% |
Knowledge Too close to call
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), o1-mini: 34.9 (#192)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| LMArena Expert | 1266 | 1316 |
| GPQA Diamond | — | 62.4% |
| Confabulations | — | 18.6% |
Multilingual o1-mini leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), o1-mini: 43.6 (#182)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| LMArena Non-English | 1273 | 1289 |
| LMArena Chinese | 1318 | 1314 |
| LMArena French | 1289 | 1293 |
| LMArena German | 1258 | 1278 |
| LMArena Japanese | 1228 | 1245 |
| LMArena Korean | 1209 | 1223 |
| LMArena Russian | 1289 | 1283 |
| LMArena Spanish | 1248 | 1303 |
Instruction Following Too close to call
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), o1-mini: 66.7 (#206)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| LMArena Instruction Following | 1280 | 1304 |
| LiveBench Instruction Following | — | 65.4% |
Long Context Too close to call
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), o1-mini: 40.1 (#161)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| LMArena Longer Query | 1301 | 1320 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), o1-mini: 48.4 (#202)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1-mini |
|---|---|---|
| LMArena Text | 1294 | 1317 |
| LMArena Creative Writing | 1285 | 1244 |
| LMArena Multi-Turn | 1297 | 1314 |
| Short-Story Creative Writing | — | 64.9% |
| LiveBench Language | — | 40.9% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than o1-mini?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 34.0 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or o1-mini better for coding?
o1-mini scores higher on coding benchmarks: 35.5 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and o1-mini share?
20 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and o1-mini has 39.