Model comparison
Qwen2.5 72B Instruct vs Qwen3 14B
Qwen3 14B is the stronger model overall, scoring 35.5 to 31.9 on the Noometry Index.
Last verified . 5 shared benchmarks.
Alibaba (Qwen)Summary
- They share 5 benchmarks with published results for both. Qwen2.5 72B Instruct scores higher in 2 categories and Qwen3 14B in 4 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3 14B leads 38.6 to 19.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 8.1% for Qwen2.5 72B Instruct and 66.4% for Qwen3 14B.
- Qwen3 14B is cheaper at $0.35 / $1.40 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
Side by side
| Qwen2.5 72B Instruct | Qwen3 14B | |
|---|---|---|
| Provider | Alibaba (Qwen) | Alibaba (Qwen) |
| Noometry Index | 31.9 | 35.5 |
| Released | 2024-09 | 2025-04 |
| Weights | Open | Open |
| Context window | 131K | 131K |
| Max output | 8K | 8K |
| Input $ / M tokens | $1.40 | $0.35 |
| Output $ / M tokens | $5.60 | $1.40 |
| Results tracked | 43 | 12 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3 14B leads
Qwen2.5 72B Instruct: 33.2 (#260), Qwen3 14B: 37.3 (#195)
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| SciCode | — | 31.6% |
| WeirdML | 16% | — |
| BigCodeBench Instruct | 45.8% | — |
| LMArena Coding | 1292 | — |
| BigCodeBench Complete | 55.9% | — |
Agentic & Tool Use Qwen3 14B leads
Qwen2.5 72B Instruct: 22.1 (#133), Qwen3 14B: 29.6 (#83)
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 41% |
| TheAgentCompany | 5.7% | — |
| BALROG | 16.2% | — |
| METR Time Horizons | 35.8% | — |
Reasoning Qwen2.5 72B Instruct leads
Qwen2.5 72B Instruct: 22.3 (#199), Qwen3 14B: 18.5 (#280)
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| DTBench | 62.9% | 64% |
| LMCA | 13.4% | 18.2% |
| Epoch Capabilities Index | 129 | 138.23 |
| Kagi LLM Benchmark | — | 49.1% |
| CritPt | — | 0% |
| Chess Puzzles | — | 4% |
| LMArena Hard Prompts | 1271 | — |
| BIG-Bench Hard | 79.8% | — |
| ForecastBench | 57.5 | — |
| HellaSwag | 84.8% | — |
| PIQA | 82.6% | — |
| WinoGrande | 82.3% | — |
Math Qwen3 14B leads
Qwen2.5 72B Instruct: 19.3 (#287), Qwen3 14B: 38.6 (#133)
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 8.1% | 66.4% |
| Omni-MATH | 33% | — |
| LMArena Math | 1283 | — |
| MATH Level 5 | 63.2% | — |
Knowledge Qwen3 14B leads
Qwen2.5 72B Instruct: 27.0 (#253), Qwen3 14B: 39.3 (#134)
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| GPQA Diamond | 49.1% | 63.8% |
| MMLU-Pro | 63.1% | — |
| Confabulations | 19.1% | — |
| Vectara Hallucination Rate | — | 5.4% |
| GPQA (HELM) | 42.6% | — |
| LMArena Expert | 1245 | — |
| ARC (AI2) Challenge | 94.5% | — |
| MMLU | 85.3% | — |
| TriviaQA | 71.9% | — |
Multilingual Not comparable
Qwen2.5 72B Instruct: 41.0 (#213), Qwen3 14B: —
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| LMArena Non-English | 1252 | — |
| LMArena Chinese | 1272 | — |
| LMArena French | 1280 | — |
| LMArena German | 1234 | — |
| LMArena Japanese | 1180 | — |
| LMArena Korean | 1188 | — |
| LMArena Russian | 1264 | — |
| LMArena Spanish | 1256 | — |
Instruction Following Not comparable
Qwen2.5 72B Instruct: 65.5 (#221), Qwen3 14B: —
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| IFEval | 80.6% | — |
| LMArena Instruction Following | 1254 | — |
Long Context Too close to call
Qwen2.5 72B Instruct: 38.9 (#188), Qwen3 14B: 38.1 (#204)
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| Fiction.LiveBench | — | 62.5% |
| LMArena Longer Query | 1282 | — |
Writing & Preference Not comparable
Qwen2.5 72B Instruct: 46.7 (#215), Qwen3 14B: —
| Benchmark | Qwen2.5 72B Instruct | Qwen3 14B |
|---|---|---|
| LMArena Text | 1269 | — |
| LMArena Creative Writing | 1221 | — |
| WildBench | 80.2% | — |
| LMArena Multi-Turn | 1272 | — |
Frequently asked questions
Is Qwen2.5 72B Instruct better than Qwen3 14B?
Qwen3 14B is the stronger model overall, scoring 35.5 to 31.9 on the Noometry Index.
Which is cheaper, Qwen2.5 72B Instruct or Qwen3 14B?
Qwen3 14B is cheaper. It lists at $0.35 per million input tokens and $1.40 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is Qwen2.5 72B Instruct or Qwen3 14B better for coding?
Qwen3 14B scores higher on coding benchmarks: 37.3 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
Both accept 131K tokens.
How many benchmarks do Qwen2.5 72B Instruct and Qwen3 14B share?
5 benchmarks have published results for both models. Qwen2.5 72B Instruct has 43 scored results on Noometry and Qwen3 14B has 12.