GPT-4.5 (Preview) — benchmark results
OpenAI's GPT-4.5 research preview, a large non-reasoning model (February 2025). Provider: OpenAI. Released 2025-02-27. Access: API.
Unified ELO 1658 ± 23, rank #351 of 1776 rated models, from 79 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Smolagents Leaderboard | 78.75 | Average Accuracy (%) | 100 |
| AI Chess Leaderboard (Continuation) | 1800 | Elo | 99.6 |
| KOFFVQA - Commonsense Reasoning | 93.56 | Score (%) | 98.8 |
| Smolagents - GAIA | 56.25 | Accuracy (%) | 98.1 |
| Smolagents - SimpleQA | 88 | Accuracy (%) | 98.1 |
| BenchTable | 78.5 | Total Score (%) | 95.8 |
| KOFFVQA - Object Attributes | 85.5 | Score (%) | 95.1 |
| KOFFVQA - Korean OCR | 95 | Score (%) | 93.8 |
| AidanBench | 3003 | Novel Answers | 92.7 |
| KOFFVQA - Table Understanding | 93.33 | Score (%) | 91.4 |
| KOFFVQA - Document Understanding | 90.67 | Score (%) | 88.9 |
| PlatinumBench (MIT) | 0.99 | Avg Error Rate (%) | 87.9 |
Interactive version: theaggregate.ai/model?slug=gpt-4-5-preview · How the rankings work · Data refreshed daily, snapshot 2026-07-22.