Model comparison
DeepSeek-V2.5 (Sep 2024) vs o1
o1 is the stronger model overall, scoring 40.9 to 37.6 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 0 categories and o1 in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in coding, where o1 leads 46.1 to 31.7.
- The biggest single-benchmark swing is Aider Polyglot: 17.8% for DeepSeek-V2.5 (Sep 2024) and 61.7% for o1.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V2.5 (Sep 2024) | o1 | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 37.6 | 40.9 |
| Released | 2024-09-06 | 2024-09-12 |
| Weights | Open | Proprietary |
| Context window | — | 200K |
| Max output | — | 100K |
| Input $ / M tokens | — | $15 |
| Output $ / M tokens | — | $60 |
| Results tracked | 22 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1 leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), o1: 46.1 (#70)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| Aider Polyglot | 17.8% | 61.7% |
| LMArena Coding | 1309 | 1367 |
| HumanEval+ | 83.5% | 89% |
| MBPP+ | 74.1% | 80.2% |
| WeirdML | — | 47.6% |
| BigCodeBench Instruct | 48.6% | — |
| LiveBench Coding | — | 69.7% |
| BigCodeBench Complete | 53.2% | — |
| CadEval | — | 56% |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, o1: 24.6 (#117)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| Cybench | — | 10% |
| METR Time Horizons | — | 51.1% |
Reasoning o1 leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), o1: 27.9 (#111)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1371 |
| SimpleBench | — | 41.7% |
| ARC-AGI-1 | — | 30.7% |
| Chess Puzzles | — | 15% |
| EnigmaEval | — | 5.7% |
| LiveBench Reasoning | — | 91.6% |
| DTBench | — | 74.7% |
| LiveBench Data Analysis | — | 65.5% |
| LMCA | — | 22.3% |
| Epoch Capabilities Index | — | 141.91 |
| LiveBench | — | 75.7% |
Math Too close to call
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), o1: 36.1 (#175)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Math | 1288 | 1388 |
| FrontierMath (Tiers 1-3) | — | 14.7% |
| OTIS Mock AIME 2024-2025 | — | 73.3% |
| LiveBench Math | — | 80.3% |
| MATH Level 5 | — | 94.7% |
| FrontierMath (Feb 2025 set) | — | 9.3% |
Knowledge o1 leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), o1: 41.5 (#110)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Expert | 1266 | 1361 |
| GPQA Diamond | — | 76.8% |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |
| Confabulations | — | 11.7% |
Multimodal Not comparable
DeepSeek-V2.5 (Sep 2024): —, o1: 34.2 (#93)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Vision | — | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |
Multilingual o1 leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), o1: 48.6 (#142)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Non-English | 1273 | 1358 |
| LMArena Chinese | 1318 | 1394 |
| LMArena French | 1289 | 1344 |
| LMArena German | 1258 | 1337 |
| LMArena Japanese | 1228 | 1346 |
| LMArena Korean | 1209 | 1396 |
| LMArena Russian | 1289 | 1356 |
| LMArena Spanish | 1248 | 1345 |
Instruction Following o1 leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), o1: 74.8 (#86)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Instruction Following | 1280 | 1367 |
| LiveBench Instruction Following | — | 81.5% |
Long Context o1 leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), o1: 50.3 (#9)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Longer Query | 1301 | 1378 |
| Fiction.LiveBench | — | 83.3% |
Writing & Preference o1 leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), o1: 55.6 (#144)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | o1 |
|---|---|---|
| LMArena Text | 1294 | 1366 |
| LMArena Creative Writing | 1285 | 1348 |
| LMArena Multi-Turn | 1297 | 1369 |
| Short-Story Creative Writing | — | 70.2% |
| LiveBench Language | — | 65.4% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than o1?
o1 is the stronger model overall, scoring 40.9 to 37.6 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or o1 better for coding?
o1 scores higher on coding benchmarks: 46.1 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and o1 share?
20 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and o1 has 52.