Model comparison
GPT-4 vs Phi-3.5-mini
GPT-4 has enough public results to be ranked (#316); Phi-3.5-mini does not yet, so treat this comparison as directional.
Last verified . 3 shared benchmarks.
MicrosoftSummary
- They share 3 benchmarks with published results for both. GPT-4 scores higher in 0 categories and Phi-3.5-mini in 1 category; one gap is clear of the uncertainty.
- The biggest single-benchmark swing is BigCodeBench Complete: 57.2% for GPT-4 and 38.5% for Phi-3.5-mini.
- Phi-3.5-mini has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Phi-3.5-mini | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 29.1 | 36.9 |
| Released | 2023-03-14 | 2024-08-16 |
| Weights | Proprietary | Open |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 5 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Phi-3.5-mini leads
GPT-4: 31.6 (#283), Phi-3.5-mini: 34.6 (#232)
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| BigCodeBench Instruct | 46% | 32.8% |
| BigCodeBench Complete | 57.2% | 38.5% |
| WeirdML | 12.4% | — |
| LMArena Coding | 1254 | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning Not comparable
GPT-4: 17.8 (#289), Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| Chess Puzzles | 4% | — |
| LMArena Hard Prompts | 1241 | — |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| Epoch Capabilities Index | 125.89 | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| PIQA | — | 81% |
| WinoGrande | 87.5% | — |
Math Not comparable
GPT-4: 10.8 (#309), Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| GSM8K | 92% | 86.2% |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| LMArena Math | 1269 | — |
| MATH Level 5 | 23% | — |
Knowledge Not comparable
GPT-4: 18.4 (#282), Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| GPQA Diamond | 35.7% | — |
| LMArena Expert | 1211 | — |
| BoolQ | — | 78% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multilingual Not comparable
GPT-4: 40.6 (#215), Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| LMArena Non-English | 1246 | — |
| LMArena Chinese | 1242 | — |
| LMArena French | 1283 | — |
| LMArena German | 1251 | — |
| LMArena Japanese | 1209 | — |
| LMArena Korean | 1184 | — |
| LMArena Russian | 1251 | — |
| LMArena Spanish | 1261 | — |
Instruction Following Not comparable
GPT-4: 65.3 (#222), Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| LMArena Instruction Following | 1241 | — |
Long Context Not comparable
GPT-4: 37.7 (#212), Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| LMArena Longer Query | 1244 | — |
Writing & Preference Not comparable
GPT-4: 34.9 (#268), Phi-3.5-mini: —
| Benchmark | GPT-4 | Phi-3.5-mini |
|---|---|---|
| LMArena Text | 1263 | — |
| LMArena Creative Writing | 1244 | — |
| EQ-Bench Creative Writing | 752 | — |
| LMArena Multi-Turn | 1257 | — |
Frequently asked questions
Is GPT-4 better than Phi-3.5-mini?
GPT-4 has enough public results to be ranked (#316); Phi-3.5-mini does not yet, so treat this comparison as directional.
Is GPT-4 or Phi-3.5-mini better for coding?
Phi-3.5-mini scores higher on coding benchmarks: 34.6 versus 31.6 in the Noometry coding category.
How many benchmarks do GPT-4 and Phi-3.5-mini share?
3 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Phi-3.5-mini has 5.