ChatGLM3-6B-Base: benchmark results
Zhipu and Tsinghua KEG's pretrained ChatGLM3-6B base checkpoint, released separately from the ChatGLM3 dialogue and long-context dialogue models. Provider: Zhipu. Access: Open.
Aggregate Rating 1395 ± 35, rank #6318 of 10916 rated models, from 35 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Portuguese LLM - FaQuAD NLI | 82.06 | Macro F1 (%) | 93.8 |
| Open PL LLM - PSC (multiple choice, 5-shot) | 71.68 | Binary F1 (%) | 62.4 |
| Open PL LLM - DYK (multiple choice, 5-shot) | 53.77 | Binary F1 (%) | 58.5 |
| Open Portuguese LLM - ASSIN2 RTE | 90.39 | Macro F1 (%) | 57 |
| C-Eval | 69 | Average (%) | 55.3 |
| Open PL LLM - PolQA Reranking (multiple choice, 5-shot) | 71.3 | Accuracy (%) | 54.4 |
| Open PL LLM - DYK (generative, 5-shot) | 52.55 | Binary F1 (%) | 53.6 |
| Open Portuguese LLM - ENEM | 61.51 | Accuracy (%) | 49.5 |
| Open PL LLM - Multiple Choice | 38.01 | Average Multiple-Choice Score (%) | 49.3 |
| Open PL LLM - PPC (multiple choice, 5-shot) | 68.5 | Accuracy (%) | 47.7 |
| Open Portuguese LLM - BLUEX | 50.21 | Accuracy (%) | 46.9 |
| Open Portuguese LLM - OAB Exams | 41.05 | Accuracy (%) | 45.7 |
Interactive version: theaggregate.ai/model?slug=chatglm3-6b-base · How It Works · Data refreshed daily, snapshot 2026-11-08.