Model comparison
Llama 3-8B vs Phi-4
Phi-4 is the stronger model overall, scoring 31.2 to 25.5 on the Noometry Index.
Last verified . 25 shared benchmarks.
Meta
MicrosoftSummary
- They share 25 benchmarks with published results for both. Llama 3-8B scores higher in 0 categories and Phi-4 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Phi-4 leads 32.6 to 7.8.
- The biggest single-benchmark swing is MATH Level 5: 6.1% for Llama 3-8B and 64.9% for Phi-4.
Side by side
| Llama 3-8B | Phi-4 | |
|---|---|---|
| Provider | Meta | Microsoft |
| Noometry Index | 25.5 | 31.2 |
| Released | 2024-04-18 | 2024-12-11 |
| Weights | Open | Open |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.07 |
| Output $ / M tokens | — | $0.14 |
| Results tracked | 34 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Phi-4 leads
Llama 3-8B: 31.0 (#289), Phi-4: 34.4 (#239)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| BigCodeBench Instruct | 31.9% | 45.5% |
| LMArena Coding | 1152 | 1231 |
| BigCodeBench Complete | 36.9% | 55.4% |
| LiveBench Coding | — | 30.7% |
| HumanEval+ | 56.7% | — |
| MBPP+ | 54.8% | — |
Agentic & Tool Use Not comparable
Llama 3-8B: —, Phi-4: 22.8 (#128)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning Phi-4 leads
Llama 3-8B: 14.3 (#326), Phi-4: 17.7 (#291)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| Chess Puzzles | 0% | 1% |
| LMArena Hard Prompts | 1133 | 1220 |
| Epoch Capabilities Index | 116.45 | 130.42 |
| LiveBench Reasoning | — | 47.8% |
| DTBench | 43.9% | — |
| LiveBench Data Analysis | — | 45.2% |
| Adversarial NLI | 57.3% | — |
| ForecastBench | 58.6 | — |
| LiveBench | — | 41.6% |
| WinoGrande | 75.7% | — |
Math Phi-4 leads
Llama 3-8B: 8.8 (#323), Phi-4: 20.8 (#285)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.9% | 13.8% |
| LMArena Math | 1151 | 1246 |
| MATH Level 5 | 6.1% | 64.9% |
| LiveBench Math | — | 42% |
Knowledge Phi-4 leads
Llama 3-8B: 7.8 (#308), Phi-4: 32.6 (#209)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| GPQA Diamond | 26.1% | 56.1% |
| LMArena Expert | 1113 | 1203 |
| MMLU | 68.8% | 84.8% |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| ARC (AI2) Challenge | 82.8% | — |
| OpenBookQA | 82.6% | — |
| TriviaQA | 67.7% | — |
Multilingual Phi-4 leads
Llama 3-8B: 30.8 (#261), Phi-4: 37.2 (#237)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| LMArena Non-English | 1098 | 1197 |
| LMArena Chinese | 1076 | 1212 |
| LMArena French | 1159 | 1224 |
| LMArena German | 1104 | 1222 |
| LMArena Japanese | 967 | 1158 |
| LMArena Korean | 1004 | 1151 |
| LMArena Russian | 1109 | 1209 |
| LMArena Spanish | 1173 | 1234 |
Instruction Following Phi-4 leads
Llama 3-8B: 58.4 (#260), Phi-4: 60.4 (#251)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1127 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context Phi-4 leads
Llama 3-8B: 34.2 (#251), Phi-4: 36.9 (#226)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1128 | 1217 |
Writing & Preference Phi-4 leads
Llama 3-8B: 37.5 (#256), Phi-4: 40.5 (#244)
| Benchmark | Llama 3-8B | Phi-4 |
|---|---|---|
| LMArena Text | 1166 | 1217 |
| LMArena Creative Writing | 1150 | 1182 |
| LMArena Multi-Turn | 1152 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is Llama 3-8B better than Phi-4?
Phi-4 is the stronger model overall, scoring 31.2 to 25.5 on the Noometry Index.
Is Llama 3-8B or Phi-4 better for coding?
Phi-4 scores higher on coding benchmarks: 34.4 versus 31.0 in the Noometry coding category.
How many benchmarks do Llama 3-8B and Phi-4 share?
25 benchmarks have published results for both models. Llama 3-8B has 34 scored results on Noometry and Phi-4 has 37.