ChatGLM 6B: benchmark results
Provider: Zhipu. Released 2023-03-14. Access: Open.
Aggregate Rating 1217 ± 31, rank #9501 of 10916 rated models, from 30 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLMsPark - Dictator Game (Single Play) | 977.77 | Elo rating (initial 1000, K=32) | 30.8 |
| LLMsPark - Prisoner's Dilemma (Single Play) | 895.12 | Elo rating (initial 1000, K=32) | 30.8 |
| Gaokao-Bench | 30.8 | Objective Overall (%) | 25 |
| LLMsPark - Dictator Game (Multiple Play) | 973.13 | Elo rating (initial 1000, K=32) | 23.1 |
| MT-Bench | 4.5 | MT-Bench average score (points) | 18.2 |
| CMMLU | 37.48 | 5-shot Avg Accuracy (%) | 15 |
| Chatbot Arena (Text - Chinese) | 1167 | Arena Score | 9.4 |
| LLMsPark - Nim Game | 880.45 | Elo rating (initial 1000, K=32) | 7.7 |
| LLMsPark - Prisoner's Dilemma (Multiple Play) | 863.98 | Elo rating (initial 1000, K=32) | 7.7 |
| C-Eval | 38.9 | Average (%) | 4.9 |
| Chatbot Arena (Text - Coding) | 1036 | Arena Score | 1.5 |
| Chatbot Arena (Text - Math) | 977 | Arena Score | 1.5 |
Interactive version: theaggregate.ai/model?slug=chatglm-6b · How It Works · Data refreshed daily, snapshot 2026-11-08.