Hunyuan-7B-Instruct: benchmark results
Tencent Hunyuan dense instruction checkpoint with grouped-query attention and switchable fast and slow thinking. The publisher supports a 256K context; pretraining checkpoints and quantized releases remain separate. Provider: Tencent. Released 2025-07-30. Access: Open.
Aggregate Rating 1324 ± 28, rank #8270 of 10916 rated models, from 271 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Portuguese NLU - ScaLA PT | 24.54 | Matthews correlation coefficient (%) | 78.6 |
| CSSBench - Adult Content | 66 | Attack success rate (%; share of obfuscated malicious adult | 77.8 |
| EuroEval Italian NLU - ScaLA IT | 28.33 | Matthews correlation coefficient (%) | 74.4 |
| CSSBench - Public and Political Safety | 35.17 | Attack success rate (%; share of obfuscated malicious public | 72.2 |
| CSSBench - Fraud and Hate | 42.67 | Attack success rate (%; share of obfuscated malicious fraud | 66.7 |
| CSSBench - Illegal Activities and Compliance | 35.33 | Attack success rate (%; share of obfuscated malicious illega | 66.7 |
| EuroEval Polish Summarization - PSC | 22.53 | chrF3++ (%) | 65.3 |
| EuroEval Bulgarian NLU - Cinexio | 34.26 | Matthews correlation coefficient (%) | 61.6 |
| EuroEval Spanish NLU - ScaLA ES | 21.95 | Matthews correlation coefficient (%) | 60.8 |
| EuroEval English NLU - ScaLA EN | 46.81 | Matthews correlation coefficient (%) | 59.9 |
| EuroEval Spanish NLU - Sentiment Headlines ES | 41.5 | Matthews correlation coefficient (%) | 57.3 |
| EuroEval Polish NLU - KPWr NER | 49.17 | Micro F1 excluding MISC (%) | 56.7 |
Interactive version: theaggregate.ai/model?slug=hunyuan-7b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.