Model comparison
GPT-3.5-turbo vs GPT-4o
GPT-4o is the stronger model overall, scoring 28.6 to 23.2 on the Noometry Index. GPT-3.5-turbo costs 5.8× less per token, which makes it the better buy when GPT-4o's lead doesn't matter for your workload.
Last verified . 33 shared benchmarks.
Summary
- They share 33 benchmarks with published results for both. GPT-3.5-turbo scores higher in 1 category and GPT-4o in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GPT-4o leads 52.6 to 25.3.
- The biggest single-benchmark swing is MATH Level 5: 15.9% for GPT-3.5-turbo and 53.3% for GPT-4o.
- GPT-3.5-turbo is cheaper at $0.50 / $1.50 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- GPT-4o accepts more context: 128K tokens versus 16K.
Side by side
| GPT-3.5-turbo | GPT-4o | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 23.2 | 28.6 |
| Released | 2023-03-01 | 2024-05-13 |
| Weights | Proprietary | Proprietary |
| Context window | 16K | 128K |
| Max output | 4K | 16K |
| Input $ / M tokens | $0.50 | $2.50 |
| Output $ / M tokens | $1.50 | $10 |
| Results tracked | 44 | 72 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
GPT-3.5-turbo: 23.9 (#331), GPT-4o: 24.8 (#328)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| WeirdML | 3.5% | 25.1% |
| BigCodeBench Instruct | 39.1% | 51.1% |
| LMArena Coding | 1136 | 1297 |
| BigCodeBench Complete | 50.6% | 61.1% |
| HumanEval+ | 70.7% | 87.2% |
| MBPP+ | 69.7% | 72.2% |
| SWE-bench Verified | — | 31% |
| SWE-bench Verified (bash only) | — | 21.6% |
| Aider Polyglot | — | 45.3% |
| GSO | — | 0% |
| LiveBench Coding | — | 51.4% |
| CadEval | — | 26% |
Agentic & Tool Use Not comparable
GPT-3.5-turbo: —, GPT-4o: 21.0 (#141)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| METR Time Horizons | 21.5% | 40.8% |
| GDPval | — | 9.9% |
| TheAgentCompany | — | 8.6% |
| Cybench | — | 12.5% |
| BALROG | — | 32.3% |
| LMArena Search | — | 1006 |
Reasoning GPT-3.5-turbo leads
GPT-3.5-turbo: 13.8 (#332), GPT-4o: 9.4 (#343)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| Chess Puzzles | 0% | 13% |
| LMArena Hard Prompts | 1108 | 1281 |
| DTBench | 48.5% | 64.5% |
| LMCA | 9.7% | 16.6% |
| Epoch Capabilities Index | 118.55 | 128.97 |
| ForecastBench | 50.4 | 57.7 |
| ARC-AGI-2 | — | 0% |
| SimpleBench | — | 17.8% |
| ARC-AGI-1 | — | 4.5% |
| CritPt | — | 0% |
| EnigmaEval | — | 0.8% |
| LiveBench Reasoning | — | 55.8% |
| Mystery Game Puzzles | 3% | — |
| LiveBench Data Analysis | — | 60.9% |
| Adversarial NLI | 58.1% | — |
| BIG-Bench Hard | 61.6% | — |
| CommonsenseQA 2.0 | 57% | — |
| LiveBench | — | 55.3% |
| WinoGrande | 81.6% | — |
Math GPT-4o leads
GPT-3.5-turbo: 6.3 (#327), GPT-4o: 10.6 (#312)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0% | 0.4% |
| OTIS Mock AIME 2024-2025 | 2.2% | 6.4% |
| LMArena Math | 1142 | 1285 |
| MATH Level 5 | 15.9% | 53.3% |
| Omni-MATH | — | 29.3% |
| LiveBench Math | — | 49.5% |
| FrontierMath (Feb 2025 set) | — | 0.3% |
| GSM8K | 57.8% | — |
Knowledge GPT-4o leads
GPT-3.5-turbo: 10.0 (#303), GPT-4o: 28.8 (#242)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| GPQA Diamond | 28% | 49.2% |
| LMArena Expert | 1070 | 1250 |
| MMLU | 71.4% | 88.1% |
| Humanity's Last Exam | — | 2.7% |
| SimpleQA Verified | — | 26% |
| MMLU-Pro | — | 71.3% |
| Confabulations | — | 15.3% |
| Vectara Hallucination Rate | — | 9.6% |
| GPQA (HELM) | — | 52% |
| ARC (AI2) Challenge | 87.4% | — |
| BoolQ | 87% | — |
| OpenBookQA | 86% | — |
| TriviaQA | 85.8% | — |
Multimodal Not comparable
GPT-3.5-turbo: —, GPT-4o: 34.5 (#91)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| LMArena Vision | — | 1137 |
| Video-MME | — | 71.9% |
| GeoBench | — | 71% |
| VPCT | — | 40% |
| ScienceQA | — | 88.5% |
Multilingual GPT-4o leads
GPT-3.5-turbo: 31.5 (#258), GPT-4o: 43.2 (#186)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| LMArena Non-English | 1108 | 1283 |
| LMArena Chinese | 1075 | 1277 |
| LMArena French | 1118 | 1304 |
| LMArena German | 1090 | 1282 |
| LMArena Japanese | 1043 | 1257 |
| LMArena Korean | 1019 | 1234 |
| LMArena Russian | 1123 | 1286 |
| LMArena Spanish | 1121 | 1292 |
Instruction Following GPT-4o leads
GPT-3.5-turbo: 57.9 (#262), GPT-4o: 66.6 (#207)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| LMArena Instruction Following | 1119 | 1278 |
| LiveBench Instruction Following | — | 68.6% |
| IFEval | — | 81.7% |
Long Context GPT-4o leads
GPT-3.5-turbo: 34.0 (#254), GPT-4o: 39.4 (#179)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| LMArena Longer Query | 1121 | 1289 |
| Fiction.LiveBench | — | 66.7% |
Writing & Preference GPT-4o leads
GPT-3.5-turbo: 25.3 (#305), GPT-4o: 52.6 (#166)
| Benchmark | GPT-3.5-turbo | GPT-4o |
|---|---|---|
| LMArena Text | 1125 | 1300 |
| LMArena Creative Writing | 1092 | 1292 |
| LMArena Multi-Turn | 1117 | 1302 |
| Short-Story Creative Writing | — | 81.8% |
| EQ-Bench Creative Writing | 451 | — |
| WildBench | — | 82.8% |
| LiveBench Language | — | 47.6% |
Frequently asked questions
Is GPT-3.5-turbo better than GPT-4o?
GPT-4o is the stronger model overall, scoring 28.6 to 23.2 on the Noometry Index. GPT-3.5-turbo costs 5.8× less per token, which makes it the better buy when GPT-4o's lead doesn't matter for your workload.
Which is cheaper, GPT-3.5-turbo or GPT-4o?
GPT-3.5-turbo is cheaper. It lists at $0.50 per million input tokens and $1.50 per million output tokens; GPT-4o lists at $2.50 and $10.
Is GPT-3.5-turbo or GPT-4o better for coding?
They score almost the same on coding (23.9 vs 24.8); test both on your own repository before choosing.
Which has the bigger context window?
GPT-4o does, with 128K tokens against 16K.
How many benchmarks do GPT-3.5-turbo and GPT-4o share?
33 benchmarks have published results for both models. GPT-3.5-turbo has 44 scored results on Noometry and GPT-4o has 72.