Atlas

Benchmarks

SAFE-Teen (106 scenarios)

Revised 106-scenario SAFE-Teen evaluation: 35 urgent-safety, 36 health-advice and 35 relationship scenarios. Branching conversations produce correlated criterion judgments. Group failure rates pool judged criteria, excluding blocked or unjudged responses, with separate no-prompt and under-18-prompt conditions. Kept separate from the historical 101-scenario set.

Models
0
Top score
—
Leader
—
Updated
—
Resolution
—
Saturation
—

No signal or difficulty estimate. This benchmark category is excluded from the capability model and its validation studies.

This is a benchmark-family page. Scores are recorded on the component benchmarks below rather than directly on the parent, because the components use different score scales and directions.

MetricModelsLeaderDirectionDate
SAFE-Teen 106: Health Advice (no prompt)16GPT-6.1 Sol4.7lower is better2026-10-12
SAFE-Teen 106: Health Advice (under-18 prompt)16Muse Spark 1.32.1lower is better2026-10-12
SAFE-Teen 106: Relationships (no prompt)16Mistral Large 4 Preview8.4lower is better2026-10-12
SAFE-Teen 106: Relationships (under-18 prompt)16Claude Opus 5.50.4lower is better2026-10-12
SAFE-Teen 106: Urgent Safety (no prompt)16GPT-6.1 Sol3.4lower is better2026-10-12
SAFE-Teen 106: Urgent Safety (under-18 prompt)16GPT-6.1 Sol2.0lower is better2026-10-12

Related benchmarks

Loading Atlas data…