Model comparison
Llama 3-8B vs Phi-2
Llama 3-8B has enough public results to be ranked (#344); Phi-2 does not yet, so treat this comparison as directional.
Last verified . 9 shared benchmarks.
Meta
MicrosoftSummary
- They share 9 benchmarks with published results for both.
Side by side
| Llama 3-8B | Phi-2 | |
|---|---|---|
| Provider | Meta | Microsoft |
| Noometry Index | 25.5 | — |
| Released | 2024-04-18 | 2023-12-12 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 34 | 11 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
Llama 3-8B: 31.0 (#289), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| HumanEval+ | 56.7% | 45.1% |
| MBPP+ | 54.8% | 54.2% |
| BigCodeBench Instruct | 31.9% | — |
| LMArena Coding | 1152 | — |
| BigCodeBench Complete | 36.9% | — |
Reasoning Not comparable
Llama 3-8B: 14.3 (#326), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| Adversarial NLI | 57.3% | 42.5% |
| Epoch Capabilities Index | 116.45 | 107.94 |
| WinoGrande | 75.7% | 54.7% |
| Chess Puzzles | 0% | — |
| LMArena Hard Prompts | 1133 | — |
| DTBench | 43.9% | — |
| BIG-Bench Hard | — | 59.4% |
| ForecastBench | 58.6 | — |
| HellaSwag | — | 53.6% |
Math Not comparable
Llama 3-8B: 8.8 (#323), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.9% | — |
| LMArena Math | 1151 | — |
| MATH Level 5 | 6.1% | — |
Knowledge Not comparable
Llama 3-8B: 7.8 (#308), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| ARC (AI2) Challenge | 82.8% | 75.9% |
| MMLU | 68.8% | 58.4% |
| OpenBookQA | 82.6% | 73.6% |
| TriviaQA | 67.7% | 45.2% |
| GPQA Diamond | 26.1% | — |
| LMArena Expert | 1113 | — |
Multilingual Not comparable
Llama 3-8B: 30.8 (#261), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| LMArena Non-English | 1098 | — |
| LMArena Chinese | 1076 | — |
| LMArena French | 1159 | — |
| LMArena German | 1104 | — |
| LMArena Japanese | 967 | — |
| LMArena Korean | 1004 | — |
| LMArena Russian | 1109 | — |
| LMArena Spanish | 1173 | — |
Instruction Following Not comparable
Llama 3-8B: 58.4 (#260), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| LMArena Instruction Following | 1127 | — |
Long Context Not comparable
Llama 3-8B: 34.2 (#251), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| LMArena Longer Query | 1128 | — |
Writing & Preference Not comparable
Llama 3-8B: 37.5 (#256), Phi-2: —
| Benchmark | Llama 3-8B | Phi-2 |
|---|---|---|
| LMArena Text | 1166 | — |
| LMArena Creative Writing | 1150 | — |
| LMArena Multi-Turn | 1152 | — |
Frequently asked questions
Is Llama 3-8B better than Phi-2?
Llama 3-8B has enough public results to be ranked (#344); Phi-2 does not yet, so treat this comparison as directional.
How many benchmarks do Llama 3-8B and Phi-2 share?
9 benchmarks have published results for both models. Llama 3-8B has 34 scored results on Noometry and Phi-2 has 11.