Model comparison
Gemma 2B vs Phi 3 Mini 4k Instruct
Gemma 2B is the stronger model overall, scoring 29.6 to 27.9 on the Noometry Index.
Last verified . 19 shared benchmarks.
Google
MicrosoftSummary
- They share 19 benchmarks with published results for both. Gemma 2B scores higher in 4 categories and Phi 3 Mini 4k Instruct in 3 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Gemma 2B leads 18.8 to 14.1.
Side by side
| Gemma 2B | Phi 3 Mini 4k Instruct | |
|---|---|---|
| Provider | Microsoft | |
| Noometry Index | 29.6 | 27.9 |
| Released | 2024-02-21 | 2024-04-23 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 23 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 2B leads
Gemma 2B: 29.4 (#305), Phi 3 Mini 4k Instruct: 26.6 (#323)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Coding | 1010 | 1093 |
| HumanEval+ | 20.7% | 59.1% |
| MBPP+ | 34.1% | 54.2% |
| LiveBench Coding | — | 15.5% |
Reasoning Gemma 2B leads
Gemma 2B: 18.8 (#275), Phi 3 Mini 4k Instruct: 14.1 (#328)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Hard Prompts | 989 | 1072 |
| BIG-Bench Hard | 35.2% | 71.7% |
| HellaSwag | 71.4% | 76.7% |
| WinoGrande | 65.4% | 70.8% |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | — | 26.8% |
| LiveBench Data Analysis | — | 34.7% |
| Adversarial NLI | — | 52.8% |
| Epoch Capabilities Index | 94.2 | — |
| LiveBench | — | 22.4% |
| PIQA | 77.3% | — |
Math Gemma 2B leads
Gemma 2B: 30.0 (#239), Phi 3 Mini 4k Instruct: 26.6 (#257)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Math | 1009 | 1111 |
| LiveBench Math | — | 15.7% |
| GSM8K | 17.7% | — |
Knowledge Not comparable
Gemma 2B: —, Phi 3 Mini 4k Instruct: 28.5 (#246)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| ARC (AI2) Challenge | 42.1% | 84.9% |
| MMLU | 42.3% | 68.8% |
| TriviaQA | 53.2% | 64% |
| LMArena Expert | — | 1045 |
| BoolQ | 69.4% | — |
| OpenBookQA | — | 88% |
Multilingual Phi 3 Mini 4k Instruct leads
Gemma 2B: 23.0 (#294), Phi 3 Mini 4k Instruct: 26.3 (#280)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Non-English | 958 | 1021 |
| LMArena Chinese | 986 | 1021 |
| LMArena Russian | 937 | 1022 |
| LMArena French | — | 1076 |
| LMArena German | — | 1044 |
| LMArena Japanese | — | 935 |
| LMArena Korean | — | 905 |
| LMArena Spanish | — | 1085 |
Instruction Following Too close to call
Gemma 2B: 48.5 (#302), Phi 3 Mini 4k Instruct: 47.7 (#303)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Instruction Following | 970 | 1053 |
| LiveBench Instruction Following | — | 39.1% |
Long Context Phi 3 Mini 4k Instruct leads
Gemma 2B: 29.9 (#291), Phi 3 Mini 4k Instruct: 31.7 (#276)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Longer Query | 981 | 1044 |
Writing & Preference Phi 3 Mini 4k Instruct leads
Gemma 2B: 24.0 (#308), Phi 3 Mini 4k Instruct: 27.6 (#300)
| Benchmark | Gemma 2B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Text | 1002 | 1073 |
| LMArena Creative Writing | 987 | 1037 |
| LMArena Multi-Turn | 945 | 1018 |
| LiveBench Language | — | 9.2% |
Frequently asked questions
Is Gemma 2B better than Phi 3 Mini 4k Instruct?
Gemma 2B is the stronger model overall, scoring 29.6 to 27.9 on the Noometry Index.
Is Gemma 2B or Phi 3 Mini 4k Instruct better for coding?
Gemma 2B scores higher on coding benchmarks: 29.4 versus 26.6 in the Noometry coding category.
How many benchmarks do Gemma 2B and Phi 3 Mini 4k Instruct share?
19 benchmarks have published results for both models. Gemma 2B has 23 scored results on Noometry and Phi 3 Mini 4k Instruct has 35.