Model comparison
Gemma 7B vs Qwen-14B
Qwen-14B is the stronger model overall, scoring 31.4 to 30.0 on the Noometry Index.
Last verified . 17 shared benchmarks.
Google
Alibaba (Qwen)Summary
- They share 17 benchmarks with published results for both. Gemma 7B scores higher in 1 category and Qwen-14B in 6 categories; one gap is clear of the uncertainty.
Side by side
| Gemma 7B | Qwen-14B | |
|---|---|---|
| Provider | Alibaba (Qwen) | |
| Noometry Index | 30.0 | 31.4 |
| Released | 2024-02-21 | 2023-09-24 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 27 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Gemma 7B: 30.5 (#294), Qwen-14B: 31.2 (#288)
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| LMArena Coding | 1048 | 1071 |
| HumanEval+ | 28.7% | — |
| MBPP+ | 43.4% | — |
Reasoning Too close to call
Gemma 7B: 19.9 (#249), Qwen-14B: 19.6 (#257)
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| LMArena Hard Prompts | 1042 | 1027 |
| BIG-Bench Hard | 55.1% | 55% |
| Epoch Capabilities Index | 111.99 | 113.03 |
| PIQA | 81.2% | 79.9% |
| Adversarial NLI | 48.7% | — |
| HellaSwag | 82.2% | — |
| LAMBADA | — | 71.1% |
| WinoGrande | 79% | — |
Math Too close to call
Gemma 7B: 31.2 (#228), Qwen-14B: 31.2 (#227)
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| LMArena Math | 1066 | 1068 |
| GSM8K | 46.4% | 61.3% |
Knowledge Not comparable
Gemma 7B: 27.3 (#252), Qwen-14B: —
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| ARC (AI2) Challenge | 78.3% | 84.4% |
| BoolQ | 83.2% | 86.2% |
| MMLU | 66.1% | 66.3% |
| LMArena Expert | 1001 | — |
| OpenBookQA | 78.6% | — |
| TriviaQA | 72.3% | — |
Multilingual Qwen-14B leads
Gemma 7B: 25.1 (#287), Qwen-14B: 27.5 (#275)
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| LMArena Non-English | 999 | 1041 |
| LMArena Chinese | 1035 | 1077 |
| LMArena French | 1025 | — |
| LMArena Russian | 993 | — |
Instruction Following Too close to call
Gemma 7B: 51.5 (#295), Qwen-14B: 52.4 (#289)
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| LMArena Instruction Following | 1017 | 1031 |
Long Context Too close to call
Gemma 7B: 31.1 (#282), Qwen-14B: 31.3 (#280)
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| LMArena Longer Query | 1022 | 1028 |
Writing & Preference Too close to call
Gemma 7B: 27.1 (#302), Qwen-14B: 27.6 (#299)
| Benchmark | Gemma 7B | Qwen-14B |
|---|---|---|
| LMArena Text | 1056 | 1051 |
| LMArena Creative Writing | 1024 | 1028 |
| LMArena Multi-Turn | 963 | 1022 |
Frequently asked questions
Is Gemma 7B better than Qwen-14B?
Qwen-14B is the stronger model overall, scoring 31.4 to 30.0 on the Noometry Index.
Is Gemma 7B or Qwen-14B better for coding?
They score almost the same on coding (30.5 vs 31.2); test both on your own repository before choosing.
How many benchmarks do Gemma 7B and Qwen-14B share?
17 benchmarks have published results for both models. Gemma 7B has 27 scored results on Noometry and Qwen-14B has 18.