OpenAI
GPT-4
The leading model of 2023. It accepts images as well as text.
- Announced
- Access
- Public
Benchmark scores
| Test | Score | Best on the tracker | How it was measured |
|---|---|---|---|
| GPQA DiamondPhD-level science questionsPhD experts: 70%Saturated | 36% | Gemini 4 Argon 96% | Independentdefault settings (gpt-4-0314)Run 23 Oct 2025 |
| OTIS Mock AIMEcompetition mathsSaturated | 0.6% | GPT-5.5 100% | Independentdefault settings (gpt-4-0314)Run 22 Oct 2025 |
| METR time horizontask length an AI agent finishes half the timeNo longer updated | 4 min | Claude Mythos Preview 17 h | IndependentMETR Horizon v1.1 run 'gpt_4'; 95% CI 1.9–8.0 min |
Also on these charts
Sources
- GPT-4OpenAI · Official · openai.com
- GPT-4 Technical Report (arXiv 2303.08774)arXiv · Paper · arxiv.org
- Epoch AI Benchmarking Hub (model release dates and runs)Epoch AI · Dataset · 9 Nov 2026 · epoch.ai