GPT-5.4 Nano (Non-reasoning): benchmark results
GPT-5.4 Nano evaluated with reasoning disabled. Provider: OpenAI. Released 2026-03-17. Access: API.
Aggregate Rating 1458 ± 17, rank #1242 of 2170 rated models, from 50 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Terminal-Bench Hard | 24.24 | Accuracy (%) | 64.5 |
| SCOPE - AFlow Coaching | 0.83 | Accuracy (0–1) | 50 |
| SCOPE - SCOPE Tile-match | 0.78 | Mean Jaccard (0–1) | 50 |
| Tau2-Bench Telecom | 34.8 | Success Rate (%) | 44.3 |
| AA TAU-2 Bench | 34.8 | Accuracy (%) | 43.2 |
| AA GPQA Diamond | 55.76 | Accuracy (%) | 29.7 |
| AA Long Context Reasoning | 29.67 | Accuracy (%) | 28.1 |
| WeirdML | 37.96 | Average Score | 28 |
| Wolfram LLM Benchmarking Project | 25.6 | Correct Functionality (%) | 23.7 |
| AA CritPt | 0 | Accuracy (%) | 20.9 |
| AA IFBench | 32.72 | Accuracy (%) | 20.3 |
| Korean CSAT 2026 (Easy Mode) - Physics I | 19 | Points (out of 50) | 19.7 |
Interactive version: theaggregate.ai/model?slug=gpt-5-4-nano-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-05.