Skip to content

Benchmark

MentalHealthBench · Behavior · Clinical reasoning and health accuracy

Scores are percentages. Higher is better.

Models and systems measured
15
Scores recorded
15
Sources
1
IQ weight
0.48

Score distribution

70.8%
Mean
70.7%
Median
6.9%
Std dev
62.1%
10th percentile
80.6%
90th percentile
61.6%
Min
61.6%82.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.48 for chat models, measured over 11 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1GPT-6 AstraOpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.82.6%
2GPT-6 LunaOpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.80.6%
3GPT-6 SolOpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.80%
4Grok 4.7OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs75.3%
5Muse Spark 1.3OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs71.1%
6Claude Opus 5.5OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs70.7%
7Claude Sonnet 5OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs66.5%
8Claude Fable 5.1OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs66.4%
9Claude Haiku 4.5OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs64.6%
10Gemini 3.8 FlashOpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs63.3%
11Gemini 3.1 Pro PreviewOpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs61.6%

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Gemini 2.5 ProOpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs62.1%
GPT-4o (March 2025)OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.65.7%
GPT-5.6 Luna (Aug 2026)OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.74.7%
GPT-5.6 Sol (Aug 2026)OpenAI evaluation · 95% CI available on source page· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.76.2%

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.