GPT-5 Mini (High): benchmark results
GPT-5 Mini evaluated at the high reasoning-effort setting. Provider: OpenAI. Released 2025-08-07. Access: API.
Aggregate Rating 1665 ± 8, rank #474 of 2168 rated models, from 299 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AJ-Bench (Agent-as-a-Judge) - Wide Search | 74.48 | F1 (%) on Wide search (WideSearch-derived item checks; 9 tas | 100 |
| LLM2014 Code 2025-09 | 75.09 | Median Score | 100 |
| LLM2014 Code 2025-09 - C++ | 7.29 | Score | 100 |
| LLM2014 Code 2025-09 - Golang | 8.4 | Score | 100 |
| LLM2014 Code 2025-09 - Java | 9 | Score | 100 |
| LLM2014 Code 2025-11 - Golang | 9.2 | Score | 100 |
| TopoBench (Easy) | 71 | Accuracy (proportion of puzzles solved, times 100; the mean | 100 |
| TopoBench (Hard) | 24 | Accuracy (proportion of puzzles solved, times 100; the mean | 100 |
| TopoBench (Medium) | 44 | Accuracy (proportion of puzzles solved, times 100; the mean | 100 |
| SLR-Bench - Easy | 100 | Accuracy (%) | 99.1 |
| SLR-Bench - Medium | 74 | Accuracy (%) | 98.2 |
| LLM2014 Logic 2025-10 | 79.37 | Median Score | 98 |
Interactive version: theaggregate.ai/model?slug=gpt-5-mini-high · How It Works · Data refreshed daily, snapshot 2026-10-20.