SuperCLUE-DSPSafeBench
Models respond to Chinese prompts that test content safety.
SuperCLUE reports scores in a public workbook.
- Made by
- SuperCLUE
- Kind
- Deception and alignment
- Level
- Everyday
- Scored
- Graded by people
- Answers
- Open-ended
- Measures
- Propensity
- Risk
- Hate, Non-violent crimes
- Behavior
- Harmful use, Bias, Lawlessness
- Language
- Chinese
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- A host
- In views
- Content safety
- Strongest model tested
-
phi-3-mini 3.8B #204 on the capability index
- Last model added
- September 2024
- Scores refreshed
- November 20th, 2026
Yi-1.5-9B-Chat-16K tops the leaderboard with 79.4 phi-3-mini 3.8B leads with 78.6 among models released by September 2024
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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 phi-3-mini 3.8B, at 78.6. This view shows only models we track, each at its best.
The newest model with a score here came out in September 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 Yi-1.5-9B-Chat-16K 79.4
- 2
phi-3-mini 3.8B 78.6
- 3 Internlm2.5-7B-Chat 77.6
- 4
MiniCPM3-4B 77.6
- 5 Qwen2.5-7B-Instruct 73.5
- 6 Mistral-7B-Instruct-v0.3 73.0
- 7 GLM-4-9B-Chat 72.4
- 8 Gemma-2-9B-it 72.3
- 9 Qwen2.5-1.5B-Instruct 71.3
- 10
Gemma 2 2B IT 70.4
- 11 Baichuan2-7B-Chat 65.3
- 12 Llama-3.1-8B-Instruct 61.5
Source: Official SuperCLUE-DSPSafeBench leaderboard
- 1
phi-3-mini 3.8B 78.6
- 2
MiniCPM3-4B 77.6
- 3 GLM-4-9B-Chat 72.4
- 4
Gemma 2 2B IT 70.4
Source: Official SuperCLUE-DSPSafeBench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Yi-1.5-9B-Chat-16K | 79.4 | Official leaderboard | ||
| 2 | | 78.6 | Apr 2024 | Official leaderboard | |
| 3 | Internlm2.5-7B-Chat | 77.6 | Official leaderboard | ||
| 4 | | 77.6 | Sep 2024 | Official leaderboard | |
| 5 | Qwen2.5-7B-Instruct | 73.5 | Official leaderboard | ||
| 6 | Mistral-7B-Instruct-v0.3 | 73.0 | Official leaderboard | ||
| 7 | GLM-4-9B-Chat | 72.4 | Jun 2024 | Official leaderboard | |
| 8 | Gemma-2-9B-it | 72.3 | Official leaderboard | ||
| 9 | Qwen2.5-1.5B-Instruct | 71.3 | Official leaderboard | ||
| 10 | | 70.4 | Jul 2024 | Official leaderboard | |
| 11 | Baichuan2-7B-Chat | 65.3 | Official leaderboard | ||
| 12 | Llama-3.1-8B-Instruct | 61.5 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 78.6 | Apr 2024 | Official leaderboard | |
| 2 | | 77.6 | Sep 2024 | Official leaderboard | |
| 3 | GLM-4-9B-Chat | 72.4 | Jun 2024 | Official leaderboard | |
| 4 | | 70.4 | Jul 2024 | Official leaderboard |
Rankings by part
We read 6 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as 总体不合规率(%), 包含违反社会主义核心价值观的内容(%) and 商业违法违规(%). Gemma 2 2B IT leads 4 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
总体不合规率(%)
4 models we track have a score here.
- 1
Gemma 2 2B IT ↑3 29.6
- 2 GLM-4-9B-Chat ↑1 27.6
- 3
MiniCPM3-4B ↓1 22.4
- 4
phi-3-mini 3.8B ↓3 21.4
包含违反社会主义核心价值观的内容(%)
4 models we track have a score here.
- 1 GLM-4-9B-Chat ↑2 25.8
- 2
phi-3-mini 3.8B ↓1 24.2
- 3
Gemma 2 2B IT ↑1 18.0
- 4
MiniCPM3-4B ↓2 17.3
商业违法违规(%)
4 models we track have a score here.
- 1
Gemma 2 2B IT ↑3 58.3
- 2 GLM-4-9B-Chat ↑1 40.0
- 3
phi-3-mini 3.8B ↓2 33.8
- 4
MiniCPM3-4B ↓2 29.0
包含歧视性内容(%)
4 models we track have a score here.
- 1
Gemma 2 2B IT ↑3 31.6
- 2 GLM-4-9B-Chat ↑1 22.9
- 3
MiniCPM3-4B ↓1 19.4
- 4
phi-3-mini 3.8B ↓3 10.0
侵犯他人合法权益(%)
4 models we track have a score here.
- 1
Gemma 2 2B IT ↑3 21.2
- 2
MiniCPM3-4B 20.4
- 3
phi-3-mini 3.8B ↓2 13.1
- 4 GLM-4-9B-Chat ↓1 0.20
无法满足特定业务类型的安全需求(%)
4 models we track have a score here.
- 1
phi-3-mini 3.8B 65.8
- 2 GLM-4-9B-Chat ↑1 53.0
- 3
MiniCPM3-4B ↓1 50.4
- 4
Gemma 2 2B IT 20.5
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official SuperCLUE-DSPSafeBench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official SuperCLUE-DSPSafeBench 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
7 of 9 checks pass, 2 warnings. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark
Catalog record 1 warning 1 pass
- Warning Task access We don’t know whether the tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (12 entries).
- Pass Models scored 4 models we track, plus 8 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in September 2024, for MiniCPM3-4B.
- Note Frontier models None of the top 10 has a score. The strongest tested is phi-3-mini 3.8B (#204).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on superclueai.com: 4 rows for models we track and 8 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.