Model comparison
DeepSeek-R1 vs Qwen2.5 72B Instruct
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 31.9 on the Noometry Index.
Last verified . 31 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 31 benchmarks with published results for both. DeepSeek-R1 scores higher in 8 categories and Qwen2.5 72B Instruct in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-R1 leads 43.8 to 19.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 66.4% for DeepSeek-R1 and 8.1% for Qwen2.5 72B Instruct.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
- DeepSeek-R1 accepts more context: 164K tokens versus 131K.
- Qwen2.5 72B Instruct has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 42.3 | 31.9 |
| Released | 2025-01-20 | 2024-09 |
| Weights | Proprietary | Open |
| Context window | 164K | 131K |
| Max output | 64K | 8K |
| Input $ / M tokens | $0.50 | $1.40 |
| Output $ / M tokens | $2.15 | $5.60 |
| Results tracked | 52 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 41.6% | 16% |
| LMArena Coding | 1427 | 1292 |
| Aider Polyglot | 71.4% | — |
| SciCode | 35.7% | — |
| BigCodeBench Instruct | — | 45.8% |
| LiveBench Coding | 66.7% | — |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| BALROG | 34.9% | 16.2% |
| METR Time Horizons | 53.8% | 35.8% |
| TheAgentCompany | — | 5.7% |
| DeepResearch Bench | 35.1% | — |
Reasoning Qwen2.5 72B Instruct leads
DeepSeek-R1: 18.6 (#278), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1416 | 1271 |
| Epoch Capabilities Index | 141.29 | 129 |
| ForecastBench | 60 | 57.5 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| Kagi LLM Benchmark | 69.4% | — |
| ARC-AGI-1 | 21.2% | — |
| CritPt | 1.1% | — |
| LiveBench Reasoning | 83.2% | — |
| DTBench | — | 62.9% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 13.4% |
| BIG-Bench Hard | — | 79.8% |
| HellaSwag | — | 84.8% |
| LiveBench | 71.6% | — |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 8.1% |
| Omni-MATH | 42.4% | 33% |
| LMArena Math | 1400 | 1283 |
| MATH Level 5 | 96.6% | 63.2% |
| LiveBench Math | 80.7% | — |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 76.3% | 49.1% |
| MMLU-Pro | 79.3% | 63.1% |
| Confabulations | 12.7% | 19.1% |
| GPQA (HELM) | 66.6% | 42.6% |
| LMArena Expert | 1394 | 1245 |
| Vectara Hallucination Rate | 11.3% | — |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1412 | 1252 |
| LMArena Chinese | 1442 | 1272 |
| LMArena French | 1417 | 1280 |
| LMArena German | 1404 | 1234 |
| LMArena Japanese | 1391 | 1180 |
| LMArena Korean | 1360 | 1188 |
| LMArena Russian | 1423 | 1264 |
| LMArena Spanish | 1411 | 1256 |
Instruction Following DeepSeek-R1 leads
DeepSeek-R1: 72.0 (#143), Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | 78.4% | 80.6% |
| LMArena Instruction Following | 1382 | 1254 |
| LiveBench Instruction Following | 80.5% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1391 | 1282 |
| Fiction.LiveBench | 75% | — |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | DeepSeek-R1 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1428 | 1269 |
| LMArena Creative Writing | 1405 | 1221 |
| WildBench | 82.8% | 80.2% |
| LMArena Multi-Turn | 1405 | 1272 |
| Short-Story Creative Writing | 83% | — |
| EQ-Bench Creative Writing | 1500 | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than Qwen2.5 72B Instruct?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 31.9 on the Noometry Index.
Which is cheaper, DeepSeek-R1 or Qwen2.5 72B Instruct?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is DeepSeek-R1 or Qwen2.5 72B Instruct better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-R1 and Qwen2.5 72B Instruct share?
31 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Qwen2.5 72B Instruct has 43.