GPT-5.4 Mini (Low): benchmark results
Provider: OpenAI. Released 2026-03-17. Access: API.
Aggregate Rating 1563 ± 19, rank #897 of 2287 rated models, from 39 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| IGG - Reveal Performed | 90.4 | Grounding accuracy (%) | 100 |
| RoboAbstention - Missing Capability | 40.9 | Abstention rate (%) on the 1067 Missing Capability instructi | 90 |
| BARISTA - Temporal VQA | 0.69 | Mean of answer accuracy and completeness, Gemini 2.5 Flash G | 75 |
| IGG - Overall | 72.4 | Grounding accuracy (%) | 75 |
| DGEval - IMDG Code MCQ | 71 | Accuracy (%) | 61.8 |
| RoboAbstention | 14.8 | Abstention rate (%) over all 6,069 instructions, image plus | 60 |
| LLM Chess (Saplin) | 176.22 | Native anchored-opponent Elo (signed points) | 54 |
| DGEval - Regulatory Recall | 13.7 | Accuracy (%) | 50 |
| RoboAbstention - Contradictory Instructions | 22 | Abstention rate (%) on the 837 Contradictory Instructions in | 50 |
| RoboAbstention - Underspecified Intent | 8.2 | Abstention rate (%) on the 1246 Underspecified Intent instru | 50 |
| ObviousBench | 90.97 | Answer pass³ (%) | 46.6 |
| LisanBench | 0.02 | Mean Path Length / Current Maximum | 45.2 |
Interactive version: theaggregate.ai/model?slug=gpt-5-4-mini-low · How It Works · Data refreshed daily, snapshot 2026-11-22.