DeceptionBench (PKU Alignment)
Models state a belief and then answer under social or strategic pressure. Judges measure how often their public answer contradicts their private reasoning.
- Made by
- Jiaming Ji and colleagues
- Released
- May 2025
- Kind
- Deception and alignment
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Propensity
- Risk
- Loss of control
- Behavior
- Harmful use
- Language
- English
- Interaction
- Multi-turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Loss of control, Deception and scheming
- Last model added
- January 2025
- Scores refreshed
- November 20th, 2026
Self-Monitor-7B tops the leaderboard with 15.6% DeepSeek-R1-Distill-Qwen-7B leads with 20.7%
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is DeepSeek-R1-Distill-Qwen-7B, at 20.7%. This view shows only models we track, each at its best.
- 1 Self-Monitor-7B 15.6%
- 2 Self-Monitor-8B 20.6%
- 3
DeepSeek-R1-Distill-Qwen-7B w STAR-1 20.7%
- 4 Llama-3.1-8B-R1-Distll 22.2%
- 5 Llama-3.1-8B-Instruct Safe CoT SFT 24.6%
- 6 Qwen2.5-7B-Instruct Safe CoT SFT 26.1%
Source: The DeceptionBench (PKU Alignment) paper
Source: The DeceptionBench (PKU Alignment) paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Self-Monitor-7B | 15.6% | Official leaderboard | ||
| 2 | Self-Monitor-8B | 20.6% | Official leaderboard | ||
| 3 | | 20.7% | w STAR-1 | Jan 2025 | Official leaderboard |
| 4 | Llama-3.1-8B-R1-Distll | 22.2% | Official leaderboard | ||
| 5 | Llama-3.1-8B-Instruct | 24.6% | Safe CoT SFT | Official leaderboard | |
| 6 | Qwen2.5-7B-Instruct | 26.1% | Safe CoT SFT | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 20.7% | w STAR-1 | Jan 2025 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The DeceptionBench (PKU Alignment) paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official DeceptionBench (PKU Alignment) leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6%
- Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts.
Gemini 4 Carbon 0%
- Covert misbehavior (Thinking Machines) In use Agentic tasks that check whether a model misbehaves and conceals it. Scored by the rate of covert misbehavior.
Inkling 2 Mini 8.30%
- Destructive actions (Thinking Machines) In use Agentic tasks that check whether a model takes destructive actions it was not authorized to take. Scored by the rate of destructive actions.
Inkling 2 Small 4.70%
- Misbehavior (Thinking Machines) In use Agentic tasks that check whether a model fabricates results, ignores instructions or hides what it did when a task is hard. Scored by the rate of misbehavior.
Inkling 2 Mini 9.80%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 22 months ago, in January 2025, for DeepSeek-R1-Distill-Qwen-7B. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the benchmark’s paper (6 entries).
- Pass Models scored 1 model we track, plus 5 other entries on its leaderboard.
- Warning Newest score The newest came 22 months ago, in January 2025, for DeepSeek-R1-Distill-Qwen-7B.
- Note Frontier models None of its models are on the capabilities index.
- Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 3 rows for models we track and 5 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.