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.
Aggregate Rating 1680 ± 10, rank #244 of 1656 rated models, from 172 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ForecastBench Baseline - Scratchpad - Dataset | 61.2 | Dataset Brier Index (difficulty-adjusted, higher is better) | 100 |
| Kagi LLM Chess Puzzles | 84.8 | Puzzles Solved (%) | 100 |
| Smolagents - Overall (Code Agent) | 78.75 | Accuracy (%) | 100 |
| Smolagents - SimpleQA (Vanilla) | 58 | Accuracy (%) | 100 |
| SuperCLUE General (March 2025) - Agent | 71.88 | Score | 100 |
| AI Chess Leaderboard (Continuation) | 1800 | Elo | 99.6 |
| KOFFVQA - Commonsense Reasoning | 93.56 | Score (%) | 98.8 |
| Smolagents - GAIA (Code Agent) | 56.25 | Accuracy (%) | 96.4 |
| Smolagents - SimpleQA (Code Agent) | 88 | Accuracy (%) | 96.4 |
| BenchTable | 78.5 | Total Score (%) | 95.8 |
| KOFFVQA - Object Attributes | 85.5 | Score (%) | 95.1 |
| ForecastBench Baseline - Zero-Shot - Market | 61.6 | Market Brier Index (difficulty-adjusted, higher is better) | 95 |
Interactive version: theaggregate.ai/model?slug=gpt-4-5-preview · How It Works · Data refreshed daily, snapshot 2026-10-20.