GPT-4.5
OpenAI·
LLMsflagshipcloud
Overview
Largest and most capable GPT model with reduced hallucinations and improved EQ.
Capabilities and innovations
Reduced hallucinationsImproved EQLarger contextUnsupervised Learning ScaleImproved Calibration
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 71.4% | unsourced |
| SWE-bench Verified | 38.3% | unsourced |
| MMLU | 90.2% | unsourced |
- Humanity’s Last Exam (no tools): not reported by the vendor
- SWE-bench Pro: benchmark did not exist at release
- MMLU-Pro: not reported by the vendor
- Terminal-Bench 2.x: benchmark did not exist at release
- MMMU-Pro: benchmark did not exist at release
Links
More from OpenAI
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.