GPT-4o is OpenAI's multimodal workhorse — comfortable switching between analyzing images and generating text without missing a beat. It handles a wide range of tasks with consistent reliability, from parsing visual content to drafting complex documents. The large 128k context window means it can hold long conversations or process lengthy documents without losing track of earlier details.
| Benchmark | Score | Type | Recorded |
|---|---|---|---|
| MT-Bench | 9.3 | GPT-4-judge | 2y ago |
| HumanEval | 90.2 | pass@1 | 2y ago |
| MATH | 76.6 | 4-shot | 2y ago |
| SciCode | 1.5 | main_problem_pass@1 | 1mo ago |
| Chatbot Arena | 1285.0 | Bradley-Terry Elo | 1y ago |
| MMLU | 88.7 | 5-shot | 2y ago |
| GPQA Diamond | 53.6 | 0-shot-CoT | 2y ago |
| SWE-Bench | 33.2 | accuracy | 1mo ago |
| HumanEval+ | 86.6 | pass@1 | 2y ago |
| Aider Polyglot | 23.1 | accuracy | 1mo ago |
| TAU2 | 42.8 | accuracy | 1mo ago |
| Humanity's Last Exam | 9.4 | 0-shot | 1y ago |
| AIME 2024 | 12.0 | accuracy | 1mo ago |