SuperCLUE-CPIF
Models follow precise instructions written in Chinese.
SuperCLUE reports scores in a public workbook.
- Made by
- SuperCLUE
- Kind
- Following precise instructions
- Level
- Everyday
- Scored
- Graded by people
- Answers
- Open-ended
- Measures
- Capability
- Language
- Chinese
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- A host
- Strongest model tested
-
Gemini 3 Pro #56 on the capability index
- Last model added
- November 2025
- Scores refreshed
- November 20th, 2026
GPT-5.1 leads with 83.6 among models released by November 2025 GPT-5.1 leads with 83.6 among models released by November 2025
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. This view shows only models we track, each at its best.
The newest model with a score here came out in November 2025, 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 GPT-5.1 high 83.6
- 2
Gemini 3 Pro 83.1
- 3 GPT-5 high 82.6
- 4 360zhinao3-o1.5 79.0
- 5 ERNIE-X1.1 75.9
- 6
DeepSeek-V3.2-Exp thinking 74.4
- 7 Claude Sonnet 4.5 reasoning 72.8
- 8 Kimi K2 Thinking 67.2
- 9
Hunyuan-T1 66.2
- 10
Gemini 2.5 Pro Jun 2025 63.6
- 11 GLM-4.6 60.8
- 12
MiniMax-M2 54.4
- 13 Grok 4 53.9
- 14
Qwen3-Max 51.8
- 15 Doubao-Seed-1.6-thinking-250715 46.7
Source: Official SuperCLUE-CPIF leaderboard
- 1 GPT-5.1 high 83.6
- 2
Gemini 3 Pro 83.1
- 3 GPT-5 high 82.6
- 4
DeepSeek-V3.2-Exp thinking 74.4
- 5 Claude Sonnet 4.5 reasoning 72.8
- 6 Kimi K2 Thinking 67.2
- 7
Hunyuan-T1 66.2
- 8
Gemini 2.5 Pro Jun 2025 63.6
- 9 GLM-4.6 60.8
- 10
MiniMax-M2 54.4
- 11 Grok 4 53.9
- 12
Qwen3-Max 51.8
Source: Official SuperCLUE-CPIF leaderboard
| # | Entry | Score | Setting | Released | Run on | Source |
|---|---|---|---|---|---|---|
| 1 | GPT-5.1 | 83.6 | high | Nov 2025 | Nov 2025 | Official leaderboard |
| 2 | | 83.1 | Nov 2025 | Nov 2025 | Official leaderboard | |
| 3 | GPT-5 | 82.6 | high | Aug 2025 | Oct 2025 | Official leaderboard |
| 4 | 360zhinao3-o1.5 | 79.0 | Nov 2025 | Official leaderboard | ||
| 5 | ERNIE-X1.1 | 75.9 | Oct 2025 | Official leaderboard | ||
| 6 | | 74.4 | thinking | Sep 2025 | Oct 2025 | Official leaderboard |
| 7 | Claude Sonnet 4.5 | 72.8 | reasoning | Sep 2025 | Oct 2025 | Official leaderboard |
| 8 | Kimi K2 Thinking | 67.2 | Nov 2025 | Nov 2025 | Official leaderboard | |
| 9 | | 66.2 | Mar 2025 | Oct 2025 | Official leaderboard | |
| 10 | | 63.6 | Jun 2025 | Oct 2025 | Official leaderboard | |
| 11 | GLM-4.6 | 60.8 | Sep 2025 | Oct 2025 | Official leaderboard | |
| 12 | | 54.4 | Oct 2025 | Nov 2025 | Official leaderboard | |
| 13 | Grok 4 | 53.9 | Jul 2025 | Oct 2025 | Official leaderboard | |
| 14 | | 51.8 | Sep 2025 | Oct 2025 | Official leaderboard | |
| 15 | Doubao-Seed-1.6-thinking-250715 | 46.7 | Oct 2025 | Official leaderboard |
| # | Model | Score | Setting | Released | Run on | Source |
|---|---|---|---|---|---|---|
| 1 | GPT-5.1 | 83.6 | high | Nov 2025 | Nov 2025 | Official leaderboard |
| 2 | | 83.1 | Nov 2025 | Nov 2025 | Official leaderboard | |
| 3 | GPT-5 | 82.6 | high | Aug 2025 | Oct 2025 | Official leaderboard |
| 4 | | 74.4 | thinking | Sep 2025 | Oct 2025 | Official leaderboard |
| 5 | Claude Sonnet 4.5 | 72.8 | reasoning | Sep 2025 | Oct 2025 | Official leaderboard |
| 6 | Kimi K2 Thinking | 67.2 | Nov 2025 | Nov 2025 | Official leaderboard | |
| 7 | | 66.2 | Mar 2025 | Oct 2025 | Official leaderboard | |
| 8 | | 63.6 | Jun 2025 | Oct 2025 | Official leaderboard | |
| 9 | GLM-4.6 | 60.8 | Sep 2025 | Oct 2025 | Official leaderboard | |
| 10 | | 54.4 | Oct 2025 | Nov 2025 | Official leaderboard | |
| 11 | Grok 4 | 53.9 | Jul 2025 | Oct 2025 | Official leaderboard | |
| 12 | | 51.8 | Sep 2025 | Oct 2025 | Official leaderboard |
Rankings by part
We read 10 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as 语义约束, 结构约束 and 量化约束. GPT-5 leads 4 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
语义约束
12 models we track have a score here.
- 1
DeepSeek-V3.2-Exp ↑3 73.3
- 2 GPT-5.1 ↓1 64.4
- 3
Gemini 2.5 Pro Jun 2025 ↑5 60.0
- 4 Grok 4 ↑7 60.0
- 5 GPT-5 ↓2 60.0
- 6
Qwen3-Max ↑6 60.0
- 7 Claude Sonnet 4.5 ↓2 60.0
- 8
Gemini 3 Pro ↓6 60.0
- 9 Kimi K2 Thinking ↓3 57.8
- 10 GLM-4.6 ↓1 55.6
结构约束
12 models we track have a score here.
- 1
Gemini 3 Pro ↑1 100.0
- 2 GPT-5.1 ↓1 97.8
- 3 Grok 4 ↑8 95.7
- 4
Hunyuan-T1 ↑3 93.5
- 5
DeepSeek-V3.2-Exp ↓1 93.5
- 6 GLM-4.6 ↑3 93.5
- 7
Gemini 2.5 Pro Jun 2025 ↑1 89.1
- 8 GPT-5 ↓5 89.1
- 9
Qwen3-Max ↑3 84.8
- 10 Claude Sonnet 4.5 ↓5 80.4
量化约束
12 models we track have a score here.
- 1 GPT-5 ↑2 100.0
- 2 GPT-5.1 ↓1 100.0
- 3 Claude Sonnet 4.5 ↑2 92.3
- 4
Gemini 3 Pro ↓2 92.3
- 5
Hunyuan-T1 ↑2 88.5
- 6
DeepSeek-V3.2-Exp ↓2 84.6
- 7 GLM-4.6 ↑2 80.8
- 8 Kimi K2 Thinking ↓2 80.8
- 9
Gemini 2.5 Pro Jun 2025 ↓1 65.4
- 10
MiniMax-M2 61.5
复合约束
12 models we track have a score here.
- 1 GPT-5 ↑2 95.6
- 2 GPT-5.1 ↓1 93.3
- 3
Gemini 3 Pro ↓1 93.3
- 4 Kimi K2 Thinking ↑2 82.2
- 5 Claude Sonnet 4.5 68.9
- 6
DeepSeek-V3.2-Exp ↓2 57.8
- 7
Gemini 2.5 Pro Jun 2025 ↑1 48.9
- 8
MiniMax-M2 ↑2 44.4
- 9
Hunyuan-T1 ↓2 42.2
- 10 GLM-4.6 ↓1 37.8
组合约束
12 models we track have a score here.
- 1 GPT-5 ↑2 72.7
- 2 Claude Sonnet 4.5 ↑3 69.7
- 3
Gemini 3 Pro ↓1 69.7
- 4
DeepSeek-V3.2-Exp 63.6
- 5 GPT-5.1 ↓4 63.6
- 6
Hunyuan-T1 ↑1 60.6
- 7
Gemini 2.5 Pro Jun 2025 ↑1 51.5
- 8 Kimi K2 Thinking ↓2 48.5
- 9 GLM-4.6 37.5
- 10
Qwen3-Max ↑2 33.3
等级1
12 models we track have a score here.
- 1 GPT-5 ↑2 97.5
- 2 GPT-5.1 ↓1 97.5
- 3
Gemini 3 Pro ↓1 97.5
- 4
Hunyuan-T1 ↑3 92.5
- 5
DeepSeek-V3.2-Exp ↓1 92.5
- 6
Gemini 2.5 Pro Jun 2025 ↑2 90.0
- 7 Claude Sonnet 4.5 ↓2 87.5
- 8 GLM-4.6 ↑1 87.5
- 9 Kimi K2 Thinking ↓3 87.5
- 10 Grok 4 ↑1 85.0
等级2
12 models we track have a score here.
- 1 GPT-5.1 94.9
- 2 GPT-5 ↑1 92.3
- 3
Gemini 3 Pro ↓1 89.7
- 4
DeepSeek-V3.2-Exp 87.2
- 5 Kimi K2 Thinking ↑1 87.2
- 6 Claude Sonnet 4.5 ↓1 82.0
- 7
Qwen3-Max ↑5 79.5
- 8 GLM-4.6 ↑1 79.5
- 9
Gemini 2.5 Pro Jun 2025 ↓1 76.9
- 10
MiniMax-M2 71.8
等级3
12 models we track have a score here.
- 1 GPT-5.1 81.6
- 2
Gemini 3 Pro 76.3
- 3 GPT-5 76.3
- 4
Hunyuan-T1 ↑3 65.8
- 5 Kimi K2 Thinking ↑1 63.2
- 6 Claude Sonnet 4.5 ↓1 63.2
- 7
DeepSeek-V3.2-Exp ↓3 63.2
- 8
Gemini 2.5 Pro Jun 2025 47.4
- 9 GLM-4.6 47.4
- 10 Grok 4 ↑1 36.8
等级4
12 models we track have a score here.
- 1
Gemini 3 Pro ↑1 87.5
- 2 GPT-5 ↑1 82.5
- 3 Claude Sonnet 4.5 ↑2 82.5
- 4 GPT-5.1 ↓3 82.5
- 5
DeepSeek-V3.2-Exp ↓1 70.0
- 6 Kimi K2 Thinking 70.0
- 7
Hunyuan-T1 67.5
- 8
Gemini 2.5 Pro Jun 2025 62.5
- 9 GLM-4.6 57.5
- 10 Grok 4 ↑1 55.0
等级5
12 models we track have a score here.
- 1
Gemini 3 Pro ↑1 63.2
- 2 GPT-5 ↑1 63.2
- 3 GPT-5.1 ↓2 60.5
- 4
DeepSeek-V3.2-Exp 57.9
- 5 Claude Sonnet 4.5 47.4
- 6
Gemini 2.5 Pro Jun 2025 ↑2 39.5
- 7
Hunyuan-T1 34.2
- 8
MiniMax-M2 ↑2 34.2
- 9 Grok 4 ↑2 31.6
- 10 GLM-4.6 ↓1 28.9
3 models have held number one on the official leaderboard since 2023
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official SuperCLUE-CPIF 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-CPIF 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 Instructions and chat
- IFEval In use Models answer about 500 prompts containing verifiable instructions such as word limits or required keywords. Code checks each instruction for strict and loose accuracy.
Gemma 4 31B IT 98.9%
- IFBench In use 58 new out-of-domain verifiable output constraints. Scored by programmatic constraint checks (strict or loose). Flint 89.2%
- MultiChallenge In use Multi-turn conversations testing instruction retention, inference memory, versioned editing and self-coherence. Graded by LLM judge with rubrics. GPT-6.1 Sol 82%
- Multi-IF In use Models follow verifiable instructions over three-turn conversations in eight languages, extending IFEval. Scored by instruction accuracy at each turn. LFM3-1.7B 89.1%
- AdvancedIF In use Models respond to over 1,600 prompts with complex, multi-turn, and system-level instructions. A judge model checks each response against expert-written rubrics.
MAI-Thinking-1 85%
- MM-IFEval In use Multimodal models follow image-grounded instructions with exact output constraints. Checks mix rule-based verification and a judge model. GPT-4o 64.6%
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 (15 entries).
- Pass Models scored 12 models we track, plus 3 other entries on its leaderboard.
- Warning Newest score The newest came 12 months ago, in November 2025, for Gemini 3 Pro.
- Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3 Pro (#56).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on superclueai.com: 12 rows for models we track and 3 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.