Model comparison
GPT-4 vs Phi 3 Mini 4k Instruct
GPT-4 is the stronger model overall, scoring 29.1 to 27.9 on the Noometry Index.
Last verified . 24 shared benchmarks.
MicrosoftSummary
- They share 24 benchmarks with published results for both. GPT-4 scores higher in 6 categories and Phi 3 Mini 4k Instruct in 2 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in instruction following, where GPT-4 leads 65.3 to 47.7.
- Phi 3 Mini 4k Instruct has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Phi 3 Mini 4k Instruct | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 29.1 | 27.9 |
| Released | 2023-03-14 | 2024-04-23 |
| Weights | Proprietary | Open |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4 leads
GPT-4: 31.6 (#283), Phi 3 Mini 4k Instruct: 26.6 (#323)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Coding | 1254 | 1093 |
| HumanEval+ | 79.3% | 59.1% |
| WeirdML | 12.4% | — |
| BigCodeBench Instruct | 46% | — |
| LiveBench Coding | — | 15.5% |
| BigCodeBench Complete | 57.2% | — |
| MBPP+ | — | 54.2% |
Agentic & Tool Use Not comparable
GPT-4: —, Phi 3 Mini 4k Instruct: —
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning GPT-4 leads
GPT-4: 17.8 (#289), Phi 3 Mini 4k Instruct: 14.1 (#328)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| Chess Puzzles | 4% | 0% |
| LMArena Hard Prompts | 1241 | 1072 |
| BIG-Bench Hard | 75.1% | 71.7% |
| HellaSwag | 95.3% | 76.7% |
| WinoGrande | 87.5% | 70.8% |
| LiveBench Reasoning | — | 26.8% |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LiveBench Data Analysis | — | 34.7% |
| LMCA | 17.1% | — |
| Adversarial NLI | — | 52.8% |
| Epoch Capabilities Index | 125.89 | — |
| ForecastBench | 57.8 | — |
| LiveBench | — | 22.4% |
Math Phi 3 Mini 4k Instruct leads
GPT-4: 10.8 (#309), Phi 3 Mini 4k Instruct: 26.6 (#257)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Math | 1269 | 1111 |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| LiveBench Math | — | 15.7% |
| MATH Level 5 | 23% | — |
| GSM8K | 92% | — |
Knowledge Phi 3 Mini 4k Instruct leads
GPT-4: 18.4 (#282), Phi 3 Mini 4k Instruct: 28.5 (#246)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Expert | 1211 | 1045 |
| MMLU | 86.4% | 68.8% |
| TriviaQA | 84.8% | 64% |
| GPQA Diamond | 35.7% | — |
| ARC (AI2) Challenge | — | 84.9% |
| OpenBookQA | — | 88% |
Multilingual GPT-4 leads
GPT-4: 40.6 (#215), Phi 3 Mini 4k Instruct: 26.3 (#280)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Non-English | 1246 | 1021 |
| LMArena Chinese | 1242 | 1021 |
| LMArena French | 1283 | 1076 |
| LMArena German | 1251 | 1044 |
| LMArena Japanese | 1209 | 935 |
| LMArena Korean | 1184 | 905 |
| LMArena Russian | 1251 | 1022 |
| LMArena Spanish | 1261 | 1085 |
Instruction Following GPT-4 leads
GPT-4: 65.3 (#222), Phi 3 Mini 4k Instruct: 47.7 (#303)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Instruction Following | 1241 | 1053 |
| LiveBench Instruction Following | — | 39.1% |
Long Context GPT-4 leads
GPT-4: 37.7 (#212), Phi 3 Mini 4k Instruct: 31.7 (#276)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Longer Query | 1244 | 1044 |
Writing & Preference GPT-4 leads
GPT-4: 34.9 (#268), Phi 3 Mini 4k Instruct: 27.6 (#300)
| Benchmark | GPT-4 | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Text | 1263 | 1073 |
| LMArena Creative Writing | 1244 | 1037 |
| LMArena Multi-Turn | 1257 | 1018 |
| EQ-Bench Creative Writing | 752 | — |
| LiveBench Language | — | 9.2% |
Frequently asked questions
Is GPT-4 better than Phi 3 Mini 4k Instruct?
GPT-4 is the stronger model overall, scoring 29.1 to 27.9 on the Noometry Index.
Is GPT-4 or Phi 3 Mini 4k Instruct better for coding?
GPT-4 scores higher on coding benchmarks: 31.6 versus 26.6 in the Noometry coding category.
How many benchmarks do GPT-4 and Phi 3 Mini 4k Instruct share?
24 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Phi 3 Mini 4k Instruct has 35.