GLM-4 Plus: benchmark results
Zhipu's flagship GLM-4-generation API model (128K context), positioned on par with GPT-4o on language understanding and long-text tasks. Provider: Zhipu. Released 2024-08-29. Access: API.
Unified ELO 1666 ± 16, rank #409 of 2656 rated models, from 222 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Language - Dialogue - Chinese (CompassBench 2409) | 67 | Score (%) | 100 |
| OpenCompass Language - Dialogue - English (CompassBench 2409) | 61.8 | Score (%) | 100 |
| OpenCompass Language - Dialogue - English (CompassBench 2411) | 68 | Score (%) | 100 |
| OpenCompass Reasoning - Common Sense - English (CompassBench 2409) | 58.2 | Score (%) | 100 |
| OpenCompass Reasoning - Humanities - Chinese (CompassBench 2409) | 66.8 | Score (%) | 100 |
| OpenCompass Reasoning - Science and Engineering - Chinese (CompassBench 2409) | 65 | Score (%) | 100 |
| OpenCompass Reasoning - Social - Chinese (CompassBench 2409) | 60.6 | Score (%) | 100 |
| OpenCompass Reasoning - Common Sense - Chinese (CompassBench 2409) | 58.2 | Score (%) | 98.3 |
| OpenCompass Reasoning - Common Sense - Chinese (CompassBench 2411) | 69.2 | Score (%) | 97 |
| OpenCompass Reasoning - Humanities - Chinese (CompassBench 2411) | 70.1 | Score (%) | 97 |
| OpenCompass Reasoning - Social - Chinese (CompassBench 2411) | 67.1 | Score (%) | 97 |
| SuperCLUE General (October 2024) - Hard | 51.09 | Score | 95.1 |
Interactive version: theaggregate.ai/model?slug=glm-4-plus · How It Works · Data refreshed daily, snapshot 2026-09-19.