CarryOnBench - Harmfulness (Benign Clarifications): leaderboard
CarryOnBench (2026) starts from seemingly harmful requests that have a benign purpose and checks whether a model gives the benign information a user needs, and how harmful its answers get, as simulated users clarify or push back. Judge-rated harmfulness from 1 to 5 in multi-turn conversations where the user clarifies the benign intent, counted over the whole conversation, taking the worst turn; lower is better.
Metric: Harmfulness (1-5): GPT-5.2 rates each response for harmful or policy-violating content (4 or more is harmful); a conversation scores its worst turn, averaged over conversations, over multi-turn conversations of 4 to 12 turns in which at least one user turn clarifies the benign intent, conversation-level, on CarryOnBench (398 seemingly harmful SORRY-Bench queries with a benign underlying intent; user turns simulated by Gemini-3-Flash); lower is better. Source: arxiv.org. Saturation forecast: Around 2029. 14 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Haiku 4.5 | 2.1 | #583 |
| 2 | Claude Sonnet 4 | 2.4 | #410 |
| 3 | GPT-5.4 | 2.7 | #135 |
| 4 | GPT-OSS-120B | 2.7 | #704 |
| 5 | GPT-5 Mini | 2.8 | #364 |
| 6 | Llama 4 Maverick | 3 | #1110 |
| 7 | OLMo 3.1 32B Instruct | 3.1 | #2445 |
| 8 | OLMo 3 7B Instruct | 3.2 | #6953 |
| 9 | Gemini 3.1 Pro (Preview) | 3.6 | #99 |
| 10 | Qwen 3 235B A22B | 3.6 | #628 |
| 11 | Qwen 3 32B | 3.9 | #972 |
| 12 | Mixtral 8x7B Instruct | 3.9 | #4696 |
| 13 | DeepSeek V3.1 | 3.9 | #511 |
| 14 | Qwen 3 Next 80B A3B Instruct | 3.9 | #778 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=carryonbench-harmfulness-benign-clarifications · How It Works · Data refreshed daily, snapshot 2026-11-21.