GPT-5 Mini (2025-08-07): benchmark results
August 7, 2025 GPT-5 Mini snapshot, tracked when sources report the dated API model. Provider: OpenAI. Released 2025-08-07. Access: API.
Aggregate Rating 1711 ± 7, rank #172 of 1656 rated models, from 1191 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CI-Repair-Bench | 18.9 | Pass@1 (%): share of the 567 real GitHub Actions CI failure | 100 |
| CI-Repair-Bench (BM25 Log Retrieval) | 10.6 | Pass@1 (%): share of the 567 real GitHub Actions CI failure | 100 |
| CI-Repair-Bench - Fault Localization | 45.33 | Top-1 localization recall (%): share of the 567 real GitHub | 100 |
| ClinSQL | 74.67 | Execution Score (%) | 100 |
| ClinSQL - Diagnostic Procedures | 75.64 | Execution Score (%) | 100 |
| ClinSQL - Patient Demographics and Admissions | 80.7 | Execution Score (%) | 100 |
| Condesion-Bench | 91.24 | Decision Satisfaction Rate (%, from a 0-1 share times 100): | 100 |
| Condesion-Bench - Condition Satisfaction Rate | 97.08 | Condition Satisfaction Rate (%, from a 0-1 share times 100): | 100 |
| Condesion-Bench - Contextual Conditions | 97.21 | Satisfaction rate (%) of contextual conditions (feasibility | 100 |
| Condesion-Bench - Variable Conditions | 94.02 | Satisfaction rate (%) of variable conditions (which decision | 100 |
| EuroEval German NLU - Sb10K | 64.7 | Matthews correlation coefficient (%) | 100 |
| EuroEval Latvian Common Sense Reasoning | 97.25 | Common Sense Reasoning Average Score (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gpt-5-mini-2025-08-07 · How It Works · Data refreshed daily, snapshot 2026-10-20.