Model comparison
GPT-4 Turbo vs Phi-2
GPT-4 Turbo has enough public results to be ranked (#292); Phi-2 does not yet, so treat this comparison as directional.
Last verified . 4 shared benchmarks.
MicrosoftSummary
- They share 4 benchmarks with published results for both.
- Phi-2 has downloadable open weights; the other is API-only.
Side by side
| GPT-4 Turbo | Phi-2 | |
|---|---|---|
| Provider | OpenAI | Microsoft |
| Noometry Index | 30.5 | — |
| Released | 2023-11-06 | 2023-12-12 |
| Weights | Proprietary | Open |
| Context window | 128K | — |
| Max output | 4K | — |
| Input $ / M tokens | $10 | — |
| Output $ / M tokens | $30 | — |
| Results tracked | 36 | 11 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
GPT-4 Turbo: 33.8 (#249), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| HumanEval+ | 86.6% | 45.1% |
| MBPP+ | 73.3% | 54.2% |
| WeirdML | 18% | — |
| BigCodeBench Instruct | 48.2% | — |
| LMArena Coding | 1268 | — |
| BigCodeBench Complete | 58.2% | — |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| METR Time Horizons | 36.7% | — |
Reasoning Not comparable
GPT-4 Turbo: 15.3 (#317), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| Epoch Capabilities Index | 127.25 | 107.94 |
| SimpleBench | 25.1% | — |
| Chess Puzzles | 6% | — |
| LMArena Hard Prompts | 1251 | — |
| DTBench | 61.6% | — |
| LMCA | 9.8% | — |
| Adversarial NLI | — | 42.5% |
| BIG-Bench Hard | — | 59.4% |
| ForecastBench | 59.4 | — |
| HellaSwag | — | 53.6% |
| WinoGrande | — | 54.7% |
Math Not comparable
GPT-4 Turbo: 9.0 (#322), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.7% | — |
| OTIS Mock AIME 2024-2025 | 6.7% | — |
| LMArena Math | 1272 | — |
| MATH Level 5 | 46.7% | — |
Knowledge Not comparable
GPT-4 Turbo: 24.3 (#268), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| MMLU | 81.3% | 58.4% |
| GPQA Diamond | 46.6% | — |
| Confabulations | 28.4% | — |
| LMArena Expert | 1223 | — |
| ARC (AI2) Challenge | — | 75.9% |
| OpenBookQA | — | 73.6% |
| TriviaQA | — | 45.2% |
Multimodal Not comparable
GPT-4 Turbo: 30.6 (#110), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| LMArena Vision | 1090 | — |
Multilingual Not comparable
GPT-4 Turbo: 40.5 (#216), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| LMArena Non-English | 1245 | — |
| LMArena Chinese | 1242 | — |
| LMArena French | 1276 | — |
| LMArena German | 1259 | — |
| LMArena Japanese | 1194 | — |
| LMArena Korean | 1187 | — |
| LMArena Russian | 1259 | — |
| LMArena Spanish | 1260 | — |
Instruction Following Not comparable
GPT-4 Turbo: 65.8 (#216), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| LMArena Instruction Following | 1249 | — |
Long Context Not comparable
GPT-4 Turbo: 38.0 (#206), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| LMArena Longer Query | 1254 | — |
Writing & Preference Not comparable
GPT-4 Turbo: 47.7 (#206), Phi-2: —
| Benchmark | GPT-4 Turbo | Phi-2 |
|---|---|---|
| LMArena Text | 1272 | — |
| LMArena Creative Writing | 1269 | — |
| LMArena Multi-Turn | 1267 | — |
Frequently asked questions
Is GPT-4 Turbo better than Phi-2?
GPT-4 Turbo has enough public results to be ranked (#292); Phi-2 does not yet, so treat this comparison as directional.
How many benchmarks do GPT-4 Turbo and Phi-2 share?
4 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and Phi-2 has 11.