Model comparison
phi-3-medium 14B vs Qwen1.5-72B
Qwen1.5-72B is the stronger model overall, scoring 30.3 to 29.1 on the Noometry Index.
Last verified . 3 shared benchmarks.
Microsoft
Alibaba (Qwen)Summary
- They share 3 benchmarks with published results for both. phi-3-medium 14B scores higher in 1 category and Qwen1.5-72B in 2 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen1.5-72B leads 30.7 to 25.2.
- The biggest single-benchmark swing is BigCodeBench Complete: 48.7% for phi-3-medium 14B and 40.3% for Qwen1.5-72B.
Side by side
| phi-3-medium 14B | Qwen1.5-72B | |
|---|---|---|
| Provider | Microsoft | Alibaba (Qwen) |
| Noometry Index | 29.1 | 30.3 |
| Released | 2024-04-23 | 2024-02-04 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 13 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding phi-3-medium 14B leads
phi-3-medium 14B: 36.3 (#198), Qwen1.5-72B: 31.4 (#272)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| BigCodeBench Instruct | 37.6% | 33.2% |
| BigCodeBench Complete | 48.7% | 40.3% |
| LMArena Coding | — | 1166 |
| HumanEval+ | — | 59.1% |
| MBPP+ | — | 61.6% |
Reasoning Not comparable
phi-3-medium 14B: —, Qwen1.5-72B: 21.2 (#201)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| LMArena Hard Prompts | — | 1148 |
| Adversarial NLI | 55.8% | — |
| BIG-Bench Hard | 81.4% | — |
| Epoch Capabilities Index | 121.18 | — |
| HellaSwag | 82.4% | — |
| WinoGrande | 81.5% | — |
Math Qwen1.5-72B leads
phi-3-medium 14B: 25.2 (#249), Qwen1.5-72B: 30.7 (#205)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| LMArena Math | — | 1164 |
| MATH Level 5 | 17.6% | — |
Knowledge Qwen1.5-72B leads
phi-3-medium 14B: 9.1 (#301), Qwen1.5-72B: 11.5 (#295)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| GPQA Diamond | 27.6% | 28.8% |
| LMArena Expert | — | 1136 |
| ARC (AI2) Challenge | 91.6% | — |
| MMLU | 78% | — |
| OpenBookQA | 87.4% | — |
| TriviaQA | 73.9% | — |
Multilingual Not comparable
phi-3-medium 14B: —, Qwen1.5-72B: 33.4 (#248)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| LMArena Non-English | — | 1135 |
| LMArena Chinese | — | 1186 |
| LMArena French | — | 1158 |
| LMArena German | — | 1084 |
| LMArena Japanese | — | 1061 |
| LMArena Korean | — | 1049 |
| LMArena Russian | — | 1104 |
| LMArena Spanish | — | 1110 |
Instruction Following Not comparable
phi-3-medium 14B: —, Qwen1.5-72B: 59.6 (#251)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| LMArena Instruction Following | — | 1142 |
Long Context Not comparable
phi-3-medium 14B: —, Qwen1.5-72B: 35.0 (#238)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| LMArena Longer Query | — | 1157 |
Writing & Preference Not comparable
phi-3-medium 14B: —, Qwen1.5-72B: 37.4 (#253)
| Benchmark | phi-3-medium 14B | Qwen1.5-72B |
|---|---|---|
| LMArena Text | — | 1166 |
| LMArena Creative Writing | — | 1137 |
| LMArena Multi-Turn | — | 1160 |
Frequently asked questions
Is phi-3-medium 14B better than Qwen1.5-72B?
Qwen1.5-72B is the stronger model overall, scoring 30.3 to 29.1 on the Noometry Index.
Is phi-3-medium 14B or Qwen1.5-72B better for coding?
phi-3-medium 14B scores higher on coding benchmarks: 36.3 versus 31.4 in the Noometry coding category.
How many benchmarks do phi-3-medium 14B and Qwen1.5-72B share?
3 benchmarks have published results for both models. phi-3-medium 14B has 13 scored results on Noometry and Qwen1.5-72B has 22.