Model comparison
GPT-3.5-turbo vs phi-3-medium 14B
phi-3-medium 14B is the stronger model overall, scoring 29.1 to 23.1 on the Noometry Index.
Last verified . 12 shared benchmarks.
MicrosoftSummary
- They share 12 benchmarks with published results for both. GPT-3.5-turbo scores higher in 1 category and phi-3-medium 14B in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in math, where phi-3-medium 14B leads 25.2 to 6.0.
- phi-3-medium 14B has downloadable open weights; the other is API-only.
Side by side
| GPT-3.5-turbo | phi-3-medium 14B | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 23.1 | 29.1 |
| Released | 2023-03-01 | 2024-04-23 |
| Weights | Proprietary | Open |
| Context window | 16K | — |
| Max output | 4K | — |
| Input $ / M tokens | $0.50 | — |
| Output $ / M tokens | $1.50 | — |
| Results tracked | 44 | 13 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding phi-3-medium 14B leads
GPT-3.5-turbo: 23.5 (#326), phi-3-medium 14B: 36.3 (#198)
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| BigCodeBench Instruct | 39.1% | 37.6% |
| BigCodeBench Complete | 50.6% | 48.7% |
| WeirdML | 3.5% | — |
| LMArena Coding | 1137 | — |
| HumanEval+ | 70.7% | — |
| MBPP+ | 69.7% | — |
Agentic & Tool Use Not comparable
GPT-3.5-turbo: —, phi-3-medium 14B: —
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| METR Time Horizons | 21.5% | — |
Reasoning Not comparable
GPT-3.5-turbo: 13.3 (#328), phi-3-medium 14B: —
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| Adversarial NLI | 58.1% | 55.8% |
| BIG-Bench Hard | 61.6% | 81.4% |
| Epoch Capabilities Index | 118.48 | 121.18 |
| WinoGrande | 81.6% | 81.5% |
| Chess Puzzles | 0% | — |
| LMArena Hard Prompts | 1108 | — |
| Mystery Game Puzzles | 3% | — |
| DTBench | 48.5% | — |
| LMCA | 9.7% | — |
| CommonsenseQA 2.0 | 57% | — |
| ForecastBench | 50.4 | — |
| HellaSwag | — | 82.4% |
Math phi-3-medium 14B leads
GPT-3.5-turbo: 6.0 (#321), phi-3-medium 14B: 25.2 (#249)
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| MATH Level 5 | 15.9% | 17.6% |
| FrontierMath (Tiers 1-3) | 0% | — |
| OTIS Mock AIME 2024-2025 | 2.2% | — |
| LMArena Math | 1142 | — |
| GSM8K | 57.8% | — |
Knowledge Too close to call
GPT-3.5-turbo: 10.0 (#298), phi-3-medium 14B: 9.1 (#301)
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| GPQA Diamond | 28% | 27.6% |
| ARC (AI2) Challenge | 87.4% | 91.6% |
| MMLU | 71.4% | 78% |
| OpenBookQA | 86% | 87.4% |
| TriviaQA | 85.8% | 73.9% |
| LMArena Expert | 1070 | — |
| BoolQ | 87% | — |
Multilingual Not comparable
GPT-3.5-turbo: 31.6 (#253), phi-3-medium 14B: —
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| LMArena Non-English | 1108 | — |
| LMArena Chinese | 1075 | — |
| LMArena French | 1117 | — |
| LMArena German | 1090 | — |
| LMArena Japanese | 1043 | — |
| LMArena Korean | 1019 | — |
| LMArena Russian | 1123 | — |
| LMArena Spanish | 1120 | — |
Instruction Following Not comparable
GPT-3.5-turbo: 58.1 (#257), phi-3-medium 14B: —
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| LMArena Instruction Following | 1119 | — |
Long Context Not comparable
GPT-3.5-turbo: 34.0 (#249), phi-3-medium 14B: —
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| LMArena Longer Query | 1121 | — |
Writing & Preference Not comparable
GPT-3.5-turbo: 25.4 (#300), phi-3-medium 14B: —
| Benchmark | GPT-3.5-turbo | phi-3-medium 14B |
|---|---|---|
| LMArena Text | 1125 | — |
| LMArena Creative Writing | 1092 | — |
| EQ-Bench Creative Writing | 451 | — |
| LMArena Multi-Turn | 1117 | — |
Frequently asked questions
Is GPT-3.5-turbo better than phi-3-medium 14B?
phi-3-medium 14B is the stronger model overall, scoring 29.1 to 23.1 on the Noometry Index.
Is GPT-3.5-turbo or phi-3-medium 14B better for coding?
phi-3-medium 14B scores higher on coding benchmarks: 36.3 versus 23.5 in the Noometry coding category.
How many benchmarks do GPT-3.5-turbo and phi-3-medium 14B share?
12 benchmarks have published results for both models. GPT-3.5-turbo has 44 scored results on Noometry and phi-3-medium 14B has 13.