Skip to content

Benchmark

Flagged-conversation factual error rate · Lower Is Better

Scores are percentages. Lower is better.

Models measured
3
Scores recorded
15
Sources
1
IQ weight
No weight published

Score distribution

5.5%
Mean
4.6%
Median
2%
Std dev
3.9%
10th percentile
7.7%
90th percentile
3.9%
Min
3.9%11.4%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1GPT-6 Astrade-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · high reasoning · displayed cost $0.48/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.3.9%
2GPT-6 Astrade-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · max reasoning · displayed cost $0.79/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.3.9%
3GPT-6 Astrade-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · xhigh reasoning · displayed cost $0.60/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4%
4GPT-6.1 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · xhigh reasoning · displayed cost $0.10/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.1%
5GPT-6 Astrade-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · medium reasoning · displayed cost $0.31/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.4%
6GPT-6 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · xhigh reasoning · displayed cost $0.13/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.5%
7GPT-6.1 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · high reasoning · displayed cost $0.08/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.5%
8GPT-6 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · max reasoning · displayed cost $0.18/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.6%
9GPT-6.1 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · max reasoning · displayed cost $0.13/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.4.6%
10GPT-6 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · high reasoning · displayed cost $0.10/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.5.1%
11GPT-6 Astrade-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · low reasoning · displayed cost $0.24/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.6.3%
12GPT-6.1 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · medium reasoning · displayed cost $0.06/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.6.3%
13GPT-6 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · medium reasoning · displayed cost $0.07/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.6.9%
14GPT-6.1 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · low reasoning · displayed cost $0.05/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.7.7%
15GPT-6 Solde-identified ChatGPT conversations flagged for prior factual errors · share of answers with any factual error · OpenAI release chart · low reasoning · displayed cost $0.05/task· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.11.4%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.