GPT-5 Mini (Minimal): benchmark results
GPT-5 Mini evaluated at the minimal reasoning-effort setting. Provider: OpenAI. Released 2025-08-07. Access: API.
Aggregate Rating 1539 ± 15, rank #922 of 2168 rated models, from 85 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| PRISM-VLM - Instruction Following | 87.5 | Per-axis mean score (%; mean compliance with 1-3 verifiable | 100 |
| PRISM-VLM - Multi-Question Robustness | 49.4 | Per-axis mean score (%; share of items whose 3-5 bundled que | 90.2 |
| PRISM-VLM | 63.3 | PScore (%; harmonic mean of the seven per-axis mean scores, | 87.5 |
| PRISM-VLM - Counting | 61.5 | Per-axis mean score (%; exact-integer accuracy on synthesize | 80.5 |
| AA Omniscience - Software Engineering (SWE) - Rust | 60 | Accuracy (%) | 80 |
| AI for Education Visual Maths - Statistics and Probability | 42.86 | Accuracy (%) | 79.7 |
| PRISM-VLM - Visual Discrimination | 86 | Per-axis mean score (%; forced A/B choice between the gold c | 77.5 |
| Mizan LLM Leaderboard | 65.6 | Average Score (0-100) | 75 |
| AA Omniscience - Software Engineering (SWE) - Go | 24 | Accuracy (%) | 74.1 |
| PRISM-VLM - Quality | 66.4 | Per-axis mean score (%; base accuracy on the recycled questi | 70.7 |
| AA Omniscience - Software Engineering (SWE) - JavaScript | 33.64 | Accuracy (%) | 69.1 |
| PRISM-VLM - Sycophancy Resistance | 58.8 | Per-axis mean score (%; share of turn-0-correct answers held | 65.9 |
Interactive version: theaggregate.ai/model?slug=gpt-5-mini-minimal · How It Works · Data refreshed daily, snapshot 2026-10-20.