GPT-5 Mini: benchmark results
OpenAI's smaller, cost-efficient GPT-5 tier. Provider: OpenAI. Released 2025-08-07. Access: API.
Unified ELO 1643 ± 1, rank #140 of 1392 rated models, from 766 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AMA-Bench - Open-World QA | 78.81 | Average Score (%) | 100 |
| AMA-Bench - Text2SQL | 83.44 | Average Score (%) | 100 |
| AMA-Bench - Web | 82.95 | Average Score (%) | 100 |
| CulturaQA | 75.09 | CulturaQA (self-reported) | 100 |
| EuroEval German NLU - Sb10K | 64.7 | Sentiment classification Score (%) | 100 |
| EuroEval Latvian Common Sense Reasoning | 97.25 | Common Sense Reasoning Average Score (%) | 100 |
| EuroEval Romanian NLU - RoSent | 98.51 | Sentiment classification Score (%) | 100 |
| HardcoreLogic - Unsolvable Puzzles | 98.52 | Accuracy (%) | 100 |
| InvisibleBench | 0 | Hard Fail Rate (self-reported) | 100 |
| MMTU - Table Understanding | 93.25 | Accuracy (%) | 100 |
| ProLLM - Summarization | 98.8 | Score (%) | 100 |
| EuroEval Latvian NLU - Latvian Twitter Sentiment | 53.12 | Sentiment classification Score (%) | 98.3 |
Interactive version: theaggregate.ai/model?slug=gpt-5-mini · How It Works · Data refreshed daily, snapshot 2026-09-05.