Model comparison
DeepSeek-V3.1-Terminus vs Qwen2.5 72B Instruct
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 42.6 to 31.5 on the Noometry Index.
Last verified . 12 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 12 benchmarks with published results for both. DeepSeek-V3.1-Terminus scores higher in 7 categories and Qwen2.5 72B Instruct in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V3.1-Terminus leads 35.7 to 18.1.
- The biggest single-benchmark swing is DTBench: 81.3% for DeepSeek-V3.1-Terminus and 62.9% for Qwen2.5 72B Instruct.
- DeepSeek-V3.1-Terminus is cheaper at $0.30 / $1 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
- DeepSeek-V3.1-Terminus accepts more context: 164K tokens versus 131K.
Side by side
| DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 42.6 | 31.5 |
| Released | 2025-09-22 | 2024-09 |
| Weights | Open | Open |
| Context window | 164K | 131K |
| Max output | 66K | 8K |
| Input $ / M tokens | $0.30 | $1.40 |
| Output $ / M tokens | $1 | $5.60 |
| Results tracked | 16 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 41.5 (#114), Qwen2.5 72B Instruct: 32.8 (#256)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Coding | 1426 | 1293 |
| SciCode | 40.6% | — |
| WeirdML | — | 16% |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 745.17 | — |
Agentic & Tool Use Not comparable
DeepSeek-V3.1-Terminus: —, Qwen2.5 72B Instruct: 22.0 (#124)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
Reasoning DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 25.7 (#126), Qwen2.5 72B Instruct: 21.4 (#197)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1427 | 1271 |
| DTBench | 81.3% | 62.9% |
| LMCA | 28.6% | 13.4% |
| Kagi LLM Benchmark | 57.4% | — |
| CritPt | 1.7% | — |
| BIG-Bench Hard | — | 79.8% |
| Epoch Capabilities Index | — | 129.01 |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 35.7 (#138), Qwen2.5 72B Instruct: 18.1 (#281)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Math | 1402 | 1283 |
| OTIS Mock AIME 2024-2025 | — | 8.1% |
| Omni-MATH | — | 33% |
| MATH Level 5 | — | 63.2% |
Knowledge Not comparable
DeepSeek-V3.1-Terminus: —, Qwen2.5 72B Instruct: 26.9 (#247)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | — | 49.1% |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| GPQA (HELM) | — | 42.6% |
| LMArena Expert | — | 1245 |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 52.3 (#87), Qwen2.5 72B Instruct: 41.2 (#208)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1407 | 1252 |
| LMArena Russian | 1434 | 1264 |
| LMArena Chinese | — | 1272 |
| LMArena French | — | 1279 |
| LMArena German | — | 1234 |
| LMArena Japanese | — | 1180 |
| LMArena Korean | — | 1188 |
| LMArena Spanish | — | 1255 |
Instruction Following DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 74.4 (#96), Qwen2.5 72B Instruct: 65.7 (#216)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Instruction Following | 1406 | 1254 |
| IFEval | — | 80.6% |
Long Context DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 43.5 (#88), Qwen2.5 72B Instruct: 38.9 (#183)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1426 | 1282 |
Writing & Preference DeepSeek-V3.1-Terminus leads
DeepSeek-V3.1-Terminus: 61.3 (#87), Qwen2.5 72B Instruct: 46.9 (#210)
| Benchmark | DeepSeek-V3.1-Terminus | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1420 | 1269 |
| LMArena Creative Writing | 1404 | 1221 |
| LMArena Multi-Turn | 1411 | 1272 |
| WildBench | — | 80.2% |
Frequently asked questions
Is DeepSeek-V3.1-Terminus better than Qwen2.5 72B Instruct?
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 42.6 to 31.5 on the Noometry Index.
Which is cheaper, DeepSeek-V3.1-Terminus or Qwen2.5 72B Instruct?
DeepSeek-V3.1-Terminus is cheaper. It lists at $0.30 per million input tokens and $1 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is DeepSeek-V3.1-Terminus or Qwen2.5 72B Instruct better for coding?
DeepSeek-V3.1-Terminus scores higher on coding benchmarks: 41.5 versus 32.8 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3.1-Terminus does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-V3.1-Terminus and Qwen2.5 72B Instruct share?
12 benchmarks have published results for both models. DeepSeek-V3.1-Terminus has 16 scored results on Noometry and Qwen2.5 72B Instruct has 43.