TIER (Qwen2.5-7B-Instruct Judge) - Explicit Harmful: leaderboard
The explicit harmful level of TIER (Trinh-Thi et al., 2026): 296 plainly harmful requests across four risk areas. Attack success rate in percent, the share of greedy responses that Qwen2.5-7B-Instruct labels harmful compliance on a six-label behavior scale; lower is better.
Metric: Attack success rate (%; L1 explicit harmful requests generated with Mistral-7B-Instruct-v0.2, 296 prompts over four risk areas; share of prompts whose response Qwen2.5-7B-Instruct labels harmful compliance (label 5 of a six-label behavior scale), greedy decoding with the default chat configuration). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | gpt-j-6B | 36.8 | #9941 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=tier-qwen2-5-7b-instruct-judge-explicit-harmful · How It Works · Data refreshed daily, snapshot 2026-11-12.