# DistressBench released to test crisis support in language models

> The new benchmark scores 25 AI models on how well they help users in suicide and self‑harm crises, with the best model meeting 88.4% of care criteria.

Oossa · 2026-10-12 · https://oossa.com/en/distressbench-released-to-test-crisis-support-in-language-models

Researchers posted DistressBench, a dataset of 718 English crisis conversations, on arXiv on 2026-10-12. Each conversation includes clinician‑written user turns and a rubric that rates seven dimensions of support. The benchmark scores models by the fraction of required care they deliver, not just whether they refuse to answer. Among 25 evaluated models, the top performer satisfied 88.4% of weighted criteria, while clinicians scored 98.7% on the same rubric.

## The facts

- 718 conversations covering 23 self‑harm subcategories
- Best model met 88.4% of care criteria; median model 71.4%

## Why it matters

The benchmark lets developers see where models fall short in real crisis help, especially in routing users to human assistance.

## Sources & references

1. [DistressBench: Evaluating Multi-Dimensional Crisis Support in Large Language Models](https://arxiv.org/abs/2610.12758) – arXiv, 2026-10-12

Last updated: 2026-10-12