LinguaGuard — original wording balanced accuracy — native trajectory-prefix prompts: leaderboard
Original-wording TSBench AgentDojo: 1,220 steps, 868 safe, 352 unsafe, 589 trajectories, 58 distinct user instructions. Balanced accuracy of supplied-action risk classification. Condition: native trajectory-prefix prompts. Managed models three repeats; self-hosted greedy single run. Appendix E explicitly identifies Qwen3.8-27B FP8 and Sonnet5 thinking disabled. Native trajectory guards predict prefix-level risk against step labels and are not pooled with step-level guards. Scanner withholds user request. Provider defaults do not resolve unknown checkpoint variants; qualified names remain distinct. FPR/TPR and translation shifts are operating-point diagnostics, not standalone general-capability scores. Metric: Balanced accuracy. Higher is better. Source: https://arxiv.org/html/2610.22244v1
Metric: Balanced accuracy. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | ShieldAgent 7B | 0.74 | |
| 2 | AgentDoG 4B | 0.66 | |
| 3 | Safiron 8B | 0.61 |
Interactive version: theaggregate.ai/benchmark?slug=linguaguard-original-wording-balanced-accuracy-native-trajectory-prefix-prompts · How It Works · Data refreshed daily, snapshot 2026-10-17.