Benchmark
ChatGPT Answers Without Errors
Scores are percentages. Higher is better.
Benchmark sourceScores from BenchBoard docs.
- Models measured
- 1
- Scores recorded
- 2
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 91%
- Mean
- 93.9%
- Median
- 3%
- Std dev
- 88%
- 10th percentile
- 93.9%
- 90th percentile
- 88%
- Min
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | GPT 5.2with search · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 93.9% |
| 2 | GPT 5.2no search · xhigh· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 88% |