Model comparison
Llama 2-70B vs Phi 3 Mini 4k Instruct
Phi 3 Mini 4k Instruct is the stronger model overall, scoring 27.4 to 24.1 on the Noometry Index.
Last verified . 24 shared benchmarks.
Meta
MicrosoftSummary
- They share 24 benchmarks with published results for both. Llama 2-70B scores higher in 6 categories and Phi 3 Mini 4k Instruct in 2 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Phi 3 Mini 4k Instruct leads 27.9 to 7.4.
Side by side
| Llama 2-70B | Phi 3 Mini 4k Instruct | |
|---|---|---|
| Provider | Meta | Microsoft |
| Noometry Index | 24.1 | 27.4 |
| Released | 2023-07-18 | 2024-04-23 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 35 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 2-70B leads
Llama 2-70B: 30.8 (#281), Phi 3 Mini 4k Instruct: 26.1 (#319)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Coding | 1079 | 1093 |
| LiveBench Coding | — | 15.5% |
| HumanEval+ | — | 59.1% |
| MBPP+ | — | 54.2% |
Reasoning Too close to call
Llama 2-70B: 13.7 (#324), Phi 3 Mini 4k Instruct: 13.6 (#326)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Hard Prompts | 1073 | 1072 |
| BIG-Bench Hard | 64.9% | 71.7% |
| HellaSwag | 85.3% | 76.7% |
| WinoGrande | 80.2% | 70.8% |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | — | 26.8% |
| DTBench | 41.6% | — |
| LiveBench Data Analysis | — | 34.7% |
| Adversarial NLI | — | 52.8% |
| CommonsenseQA 2.0 | 50% | — |
| Epoch Capabilities Index | 113.59 | — |
| ForecastBench | 51.4 | — |
| LAMBADA | 78.9% | — |
| LiveBench | — | 22.4% |
| PIQA | 82.8% | — |
Math Phi 3 Mini 4k Instruct leads
Llama 2-70B: 7.5 (#320), Phi 3 Mini 4k Instruct: 24.5 (#258)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Math | 1091 | 1111 |
| OTIS Mock AIME 2024-2025 | 0% | — |
| LiveBench Math | — | 15.7% |
| MATH Level 5 | 3.3% | — |
| GSM8K | 69.6% | — |
Knowledge Phi 3 Mini 4k Instruct leads
Llama 2-70B: 7.4 (#305), Phi 3 Mini 4k Instruct: 27.9 (#241)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Expert | 1039 | 1045 |
| ARC (AI2) Challenge | 78.3% | 84.9% |
| MMLU | 69.9% | 68.8% |
| OpenBookQA | 60.2% | 88% |
| TriviaQA | 87.6% | 64% |
| GPQA Diamond | 26.3% | — |
| BoolQ | 88.6% | — |
Multilingual Llama 2-70B leads
Llama 2-70B: 27.9 (#269), Phi 3 Mini 4k Instruct: 26.5 (#275)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Non-English | 1045 | 1021 |
| LMArena Chinese | 994 | 1021 |
| LMArena French | 1089 | 1075 |
| LMArena German | 1041 | 1044 |
| LMArena Japanese | 927 | 935 |
| LMArena Korean | 964 | 905 |
| LMArena Russian | 1083 | 1022 |
| LMArena Spanish | 1142 | 1084 |
Instruction Following Llama 2-70B leads
Llama 2-70B: 55.1 (#273), Phi 3 Mini 4k Instruct: 48.0 (#298)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Instruction Following | 1071 | 1053 |
| LiveBench Instruction Following | — | 39.1% |
Long Context Too close to call
Llama 2-70B: 32.2 (#265), Phi 3 Mini 4k Instruct: 31.7 (#271)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Longer Query | 1062 | 1044 |
Writing & Preference Llama 2-70B leads
Llama 2-70B: 32.5 (#274), Phi 3 Mini 4k Instruct: 27.7 (#295)
| Benchmark | Llama 2-70B | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Text | 1115 | 1073 |
| LMArena Creative Writing | 1075 | 1037 |
| LMArena Multi-Turn | 1088 | 1018 |
| LiveBench Language | — | 9.2% |
Frequently asked questions
Is Llama 2-70B better than Phi 3 Mini 4k Instruct?
Phi 3 Mini 4k Instruct is the stronger model overall, scoring 27.4 to 24.1 on the Noometry Index.
Is Llama 2-70B or Phi 3 Mini 4k Instruct better for coding?
Llama 2-70B scores higher on coding benchmarks: 30.8 versus 26.1 in the Noometry coding category.
How many benchmarks do Llama 2-70B and Phi 3 Mini 4k Instruct share?
24 benchmarks have published results for both models. Llama 2-70B has 35 scored results on Noometry and Phi 3 Mini 4k Instruct has 35.