Model comparison
GPT-3.5-turbo vs Phi 3 Mini 4k Instruct
Phi 3 Mini 4k Instruct is the stronger model overall, scoring 27.4 to 23.1 on the Noometry Index.
Last verified . 27 shared benchmarks.
MicrosoftSummary
- They share 27 benchmarks with published results for both. GPT-3.5-turbo scores higher in 3 categories and Phi 3 Mini 4k Instruct in 5 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Phi 3 Mini 4k Instruct leads 24.5 to 6.0.
- Phi 3 Mini 4k Instruct has downloadable open weights; the other is API-only.
Side by side
| GPT-3.5-turbo | Phi 3 Mini 4k Instruct | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 23.1 | 27.4 |
| 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 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Phi 3 Mini 4k Instruct leads
GPT-3.5-turbo: 23.5 (#326), Phi 3 Mini 4k Instruct: 26.1 (#319)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Coding | 1137 | 1093 |
| HumanEval+ | 70.7% | 59.1% |
| MBPP+ | 69.7% | 54.2% |
| WeirdML | 3.5% | — |
| BigCodeBench Instruct | 39.1% | — |
| LiveBench Coding | — | 15.5% |
| BigCodeBench Complete | 50.6% | — |
Agentic & Tool Use Not comparable
GPT-3.5-turbo: —, Phi 3 Mini 4k Instruct: —
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| METR Time Horizons | 21.5% | — |
Reasoning Too close to call
GPT-3.5-turbo: 13.3 (#328), Phi 3 Mini 4k Instruct: 13.6 (#326)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| Chess Puzzles | 0% | 0% |
| LMArena Hard Prompts | 1108 | 1072 |
| Adversarial NLI | 58.1% | 52.8% |
| BIG-Bench Hard | 61.6% | 71.7% |
| WinoGrande | 81.6% | 70.8% |
| LiveBench Reasoning | — | 26.8% |
| Mystery Game Puzzles | 3% | — |
| DTBench | 48.5% | — |
| LiveBench Data Analysis | — | 34.7% |
| LMCA | 9.7% | — |
| CommonsenseQA 2.0 | 57% | — |
| Epoch Capabilities Index | 118.48 | — |
| ForecastBench | 50.4 | — |
| HellaSwag | — | 76.7% |
| LiveBench | — | 22.4% |
Math Phi 3 Mini 4k Instruct leads
GPT-3.5-turbo: 6.0 (#321), Phi 3 Mini 4k Instruct: 24.5 (#258)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Math | 1142 | 1111 |
| FrontierMath (Tiers 1-3) | 0% | — |
| OTIS Mock AIME 2024-2025 | 2.2% | — |
| LiveBench Math | — | 15.7% |
| MATH Level 5 | 15.9% | — |
| GSM8K | 57.8% | — |
Knowledge Phi 3 Mini 4k Instruct leads
GPT-3.5-turbo: 10.0 (#298), Phi 3 Mini 4k Instruct: 27.9 (#241)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Expert | 1070 | 1045 |
| ARC (AI2) Challenge | 87.4% | 84.9% |
| MMLU | 71.4% | 68.8% |
| OpenBookQA | 86% | 88% |
| TriviaQA | 85.8% | 64% |
| GPQA Diamond | 28% | — |
| BoolQ | 87% | — |
Multilingual GPT-3.5-turbo leads
GPT-3.5-turbo: 31.6 (#253), Phi 3 Mini 4k Instruct: 26.5 (#275)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Non-English | 1108 | 1021 |
| LMArena Chinese | 1075 | 1021 |
| LMArena French | 1117 | 1075 |
| LMArena German | 1090 | 1044 |
| LMArena Japanese | 1043 | 935 |
| LMArena Korean | 1019 | 905 |
| LMArena Russian | 1123 | 1022 |
| LMArena Spanish | 1120 | 1084 |
Instruction Following GPT-3.5-turbo leads
GPT-3.5-turbo: 58.1 (#257), Phi 3 Mini 4k Instruct: 48.0 (#298)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Instruction Following | 1119 | 1053 |
| LiveBench Instruction Following | — | 39.1% |
Long Context GPT-3.5-turbo leads
GPT-3.5-turbo: 34.0 (#249), Phi 3 Mini 4k Instruct: 31.7 (#271)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Longer Query | 1121 | 1044 |
Writing & Preference Phi 3 Mini 4k Instruct leads
GPT-3.5-turbo: 25.4 (#300), Phi 3 Mini 4k Instruct: 27.7 (#295)
| Benchmark | GPT-3.5-turbo | Phi 3 Mini 4k Instruct |
|---|---|---|
| LMArena Text | 1125 | 1073 |
| LMArena Creative Writing | 1092 | 1037 |
| LMArena Multi-Turn | 1117 | 1018 |
| EQ-Bench Creative Writing | 451 | — |
| LiveBench Language | — | 9.2% |
Frequently asked questions
Is GPT-3.5-turbo better than Phi 3 Mini 4k Instruct?
Phi 3 Mini 4k Instruct is the stronger model overall, scoring 27.4 to 23.1 on the Noometry Index.
Is GPT-3.5-turbo or Phi 3 Mini 4k Instruct better for coding?
Phi 3 Mini 4k Instruct scores higher on coding benchmarks: 26.1 versus 23.5 in the Noometry coding category.
How many benchmarks do GPT-3.5-turbo and Phi 3 Mini 4k Instruct share?
27 benchmarks have published results for both models. GPT-3.5-turbo has 44 scored results on Noometry and Phi 3 Mini 4k Instruct has 35.