Model comparison
Claude 3 Opus vs DeepSeek-V2.5 (Sep 2024)
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 29.5 on the Noometry Index.
Last verified . 21 shared benchmarks.
DeepSeekSummary
- They share 21 benchmarks with published results for both. Claude 3 Opus scores higher in 1 category and DeepSeek-V2.5 (Sep 2024) in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V2.5 (Sep 2024) leads 35.9 to 14.8.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 29.5 | 37.6 |
| Released | 2024-02-29 | 2024-09-06 |
| Weights | Proprietary | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 46 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude 3 Opus leads
Claude 3 Opus: 32.9 (#267), DeepSeek-V2.5 (Sep 2024): 31.7 (#281)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| BigCodeBench Instruct | 45.5% | 48.6% |
| LMArena Coding | 1264 | 1309 |
| BigCodeBench Complete | 57.4% | 53.2% |
| HumanEval+ | 77.4% | 83.5% |
| MBPP+ | 73.3% | 74.1% |
| Aider Polyglot | — | 17.8% |
| WeirdML | 19.2% | — |
| LiveBench Coding | 38.6% | — |
Agentic & Tool Use Not comparable
Claude 3 Opus: 24.6 (#116), DeepSeek-V2.5 (Sep 2024): —
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| Cybench | 10% | — |
| METR Time Horizons | 29.5% | — |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
Claude 3 Opus: 14.6 (#324), DeepSeek-V2.5 (Sep 2024): 25.6 (#145)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Hard Prompts | 1245 | 1289 |
| SimpleBench | 23.5% | — |
| Chess Puzzles | 5% | — |
| EnigmaEval | 0.8% | — |
| LiveBench Reasoning | 40.6% | — |
| DTBench | 61.6% | — |
| LiveBench Data Analysis | 57.9% | — |
| LMCA | 17% | — |
| Epoch Capabilities Index | 126.91 | — |
| ForecastBench | 58.4 | — |
| LiveBench | 49.2% | — |
| WinoGrande | 88.5% | — |
Math DeepSeek-V2.5 (Sep 2024) leads
Claude 3 Opus: 14.8 (#299), DeepSeek-V2.5 (Sep 2024): 35.9 (#177)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Math | 1273 | 1288 |
| OTIS Mock AIME 2024-2025 | 4.7% | — |
| LiveBench Math | 43.6% | — |
| MATH Level 5 | 37.5% | — |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
Claude 3 Opus: 24.5 (#267), DeepSeek-V2.5 (Sep 2024): 34.8 (#193)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Expert | 1223 | 1266 |
| GPQA Diamond | 47.2% | — |
| SimpleQA Verified | 12.6% | — |
| Confabulations | 22.7% | — |
| MMLU | 84.6% | — |
Multimodal Not comparable
Claude 3 Opus: 27.1 (#116), DeepSeek-V2.5 (Sep 2024): —
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Vision | 1023 | — |
Multilingual DeepSeek-V2.5 (Sep 2024) leads
Claude 3 Opus: 41.4 (#207), DeepSeek-V2.5 (Sep 2024): 42.5 (#193)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Non-English | 1258 | 1273 |
| LMArena Chinese | 1248 | 1318 |
| LMArena French | 1275 | 1289 |
| LMArena German | 1258 | 1258 |
| LMArena Japanese | 1204 | 1228 |
| LMArena Korean | 1187 | 1209 |
| LMArena Russian | 1280 | 1289 |
| LMArena Spanish | 1246 | 1248 |
Instruction Following DeepSeek-V2.5 (Sep 2024) leads
Claude 3 Opus: 64.1 (#228), DeepSeek-V2.5 (Sep 2024): 67.5 (#194)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Instruction Following | 1248 | 1280 |
| LiveBench Instruction Following | 63.9% | — |
Long Context DeepSeek-V2.5 (Sep 2024) leads
Claude 3 Opus: 38.2 (#202), DeepSeek-V2.5 (Sep 2024): 39.5 (#174)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Longer Query | 1259 | 1301 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
Claude 3 Opus: 47.2 (#213), DeepSeek-V2.5 (Sep 2024): 49.8 (#187)
| Benchmark | Claude 3 Opus | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Text | 1262 | 1294 |
| LMArena Creative Writing | 1235 | 1285 |
| LMArena Multi-Turn | 1275 | 1297 |
| LiveBench Language | 50.4% | — |
Frequently asked questions
Is Claude 3 Opus better than DeepSeek-V2.5 (Sep 2024)?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 29.5 on the Noometry Index.
Is Claude 3 Opus or DeepSeek-V2.5 (Sep 2024) better for coding?
Claude 3 Opus scores higher on coding benchmarks: 32.9 versus 31.7 in the Noometry coding category.
How many benchmarks do Claude 3 Opus and DeepSeek-V2.5 (Sep 2024) share?
21 benchmarks have published results for both models. Claude 3 Opus has 46 scored results on Noometry and DeepSeek-V2.5 (Sep 2024) has 22.