Model comparison
Phi-2 vs Phi 3 Small 8k Instruct
Phi 3 Small 8k Instruct has enough public results to be ranked (#314); Phi-2 does not yet, so treat this comparison as directional.
Last verified . 8 shared benchmarks.
MicrosoftSummary
- They share 8 benchmarks with published results for both.
Side by side
| Phi-2 | Phi 3 Small 8k Instruct | |
|---|---|---|
| Provider | Microsoft | Microsoft |
| Noometry Index | — | 29.3 |
| Released | 2023-12-12 | 2024-04-23 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 11 | 32 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 27.9 (#318)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| LiveBench Coding | — | 20.3% |
| LMArena Coding | — | 1101 |
| HumanEval+ | 45.1% | — |
| MBPP+ | 54.2% | — |
Reasoning Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 14.8 (#323)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| Adversarial NLI | 42.5% | 58.1% |
| BIG-Bench Hard | 59.4% | 79.1% |
| HellaSwag | 53.6% | 77% |
| WinoGrande | 54.7% | 81.5% |
| LiveBench Reasoning | — | 15.9% |
| LMArena Hard Prompts | — | 1100 |
| LiveBench Data Analysis | — | 30.3% |
| Epoch Capabilities Index | 107.94 | — |
| LiveBench | — | 24% |
Math Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 27.6 (#248)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| LiveBench Math | — | 17.6% |
| LMArena Math | — | 1151 |
Knowledge Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 29.1 (#240)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| ARC (AI2) Challenge | 75.9% | 90.7% |
| MMLU | 58.4% | 75.7% |
| OpenBookQA | 73.6% | 88% |
| TriviaQA | 45.2% | 58.1% |
| LMArena Expert | — | 1067 |
Multilingual Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 28.5 (#272)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Non-English | — | 1058 |
| LMArena Chinese | — | 1061 |
| LMArena French | — | 1135 |
| LMArena German | — | 1080 |
| LMArena Japanese | — | 966 |
| LMArena Korean | — | 894 |
| LMArena Russian | — | 1111 |
| LMArena Spanish | — | 1111 |
Instruction Following Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 51.9 (#292)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| LiveBench Instruction Following | — | 47.2% |
| LMArena Instruction Following | — | 1087 |
Long Context Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 33.0 (#267)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Longer Query | — | 1088 |
Writing & Preference Not comparable
Phi-2: —, Phi 3 Small 8k Instruct: 31.1 (#284)
| Benchmark | Phi-2 | Phi 3 Small 8k Instruct |
|---|---|---|
| LMArena Text | — | 1110 |
| LMArena Creative Writing | — | 1083 |
| LMArena Multi-Turn | — | 1068 |
| LiveBench Language | — | 12.9% |
Frequently asked questions
Is Phi-2 better than Phi 3 Small 8k Instruct?
Phi 3 Small 8k Instruct has enough public results to be ranked (#314); Phi-2 does not yet, so treat this comparison as directional.
How many benchmarks do Phi-2 and Phi 3 Small 8k Instruct share?
8 benchmarks have published results for both models. Phi-2 has 11 scored results on Noometry and Phi 3 Small 8k Instruct has 32.