NUWA Evidence record
Qwen3.6-Max (preview)
StrongREJECT · Evidence record
Source snapshot, chart and review scope →
Original source charts and materials
Recorded measurements
- Baseline StrongREJECT
- 0.02
Source denominator: 313 (unit not disclosed)
Harness not specified · Budget not specified
Method and scope
A graded response evaluator for jailbreak effectiveness.
- Evidence type
- Benchmark
- Assessment subject
- Model responses graded for jailbreak effectiveness
- Primary dimension
- Behavior
- Evidence snapshot
- 2026-10-29.1
General safety evidence retained for context; no defensible direct link to the six NUWA risk entries was found.
Source finding
Baseline StrongREJECT: 0.02 score. Conditions are retained from the source.
Limitations
A response score is not an end-to-end measure of harmful action.
StrongREJECT aggregate score scale is not specified in the table.
Source retained for methodological reading. A reliable result figure was not obtained in this audit; no numerical score is fabricated.
Related research questions
References
- [1] Alexandra Souly ∗, Qingyuan Lu ∗, et al.. A StrongREJECT for Empty Jailbreaks. 2024-08-27. v2 [arXiv]; Author affiliations on the original paper title page; Figure 2
- [2] Evaluating DeepSeek v4 Pro for Frontier Risks. 2026-05-29. v1.1 · 29 May 2026 ; PDF p. 38; StrongREJECT; Baseline StrongREJECT