Rankings / OpenAI
GPT 5.4 Pro (2026-03-05)
released 2026-03-05
BenchAtlas Index
as of 2026-07-14
base configuration
rank #28
4 families · 4 categories · medium
Benchmark evidence
19 results
external indices
| ECI | xhigh effort | 157.8 points | independent | Epoch AI Benchmarking Hub |
| ECI | no reasoning | 157.8 points | independent | Epoch AI Benchmarking Hub |
| ECI | 157.8 points | independent | Epoch AI Benchmarking Hub |
factuality
| MASK contamination=Potential contamination warning: This model was evaluated after the public release of MASK, allowing model builder access to the prompts and solutions. | 91.7 % 89.7–93.7 | independent | Scale Labs | |
| SimpleQA Verified | xhigh effort | 47.8 % | independent | Epoch AI Benchmarking Hub 2026-03-19 |
knowledge science
| GPQA Diamond | xhigh effort | 94.6 % | independent | Epoch AI Benchmarking Hub 2026-03-20 |
| Humanity's Last Exam contamination=Potential contamination warning: This model was evaluated after the public release of HLE, allowing model builder access to the prompts and solutions. · implementation=scale | 44.3 % 42.4–46.3 | independent | Scale Labs |
long context instruction
| MultiChallenge | 69.2 % 66.2–72.3 | independent | Scale Labs |
multimodal
| VISTA | 53.9 % 51.9–55.9 | independent | Scale Labs |
reasoning math
| ARC-AGI-1older version split=public_eval · model_type=CoT | xhigh effort | 98.3 % | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | xhigh effort | 94.5 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | xhigh effort | 92.2 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | xhigh effort | 83.3 % | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=public_eval · model_type=CoT | xhigh effort | 17.6 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-2 split=semi_private · model_type=CoT | xhigh effort | 16.4 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=semi_private · model_type=CoT | xhigh effort | 7.2 usd_per_task | independent | ARC Prize Leaderboard |
| ARC-AGI-1older version split=public_eval · model_type=CoT | xhigh effort | 5.4 usd_per_task | independent | ARC Prize Leaderboard |
| EnigmaEval contamination=Potential contamination warning: This model was evaluated after the public release of EnigmaEval, allowing model builder access to the prompts and solutions. | 23.8 % 21.4–26.3 | independent | Scale Labs | |
| FrontierMath | xhigh effort | 50.0 % | independent | Epoch AI Benchmarking Hub 2026-03-06 |