GLM-4.6V: benchmark results
Zhipu's open vision-language MoE (106B total/12B active) with native tool calling and 128K context (December 2025). Provider: Zhipu. Released 2025-12-09. Access: Open.
Unified ELO 1579 ± 1, rank #323 of 1392 rated models, from 77 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MechVQA | 78.91 | Total (self-reported) | 93.8 |
| LithoBench | 94.5 | MCQ$_{\mathrm{all}}$ (self-reported) | 83.3 |
| Ko-AgentBench - L5 Error Handling & Robustness | 26.01 | Adaptive Routing Score (%) | 78.6 |
| MathVision | 63.5 | Overall Accuracy (%) | 76.9 |
| SnakeBench | 26.1 | TrueSkill Rating | 72.6 |
| CULTURE-MT | 16.07 | Ineff. (%) (self-reported) | 71.4 |
| AI for Education Visual Maths - Number and Operations | 59.46 | Accuracy (%) | 65.5 |
| Chatbot Arena (Text - Chinese) | 1429 | Arena Score | 60.7 |
| AI for Education Visual Reasoning - odd one out | 53.2 | Accuracy (%) | 60 |
| Chatbot Arena (Text - Creative Writing) | 1349 | Arena Score | 59.1 |
| LongWebBench | 5.96 | Overall (self-reported) | 58.3 |
| AI for Education Visual Maths - Geometry | 54.62 | Accuracy (%) | 58.1 |
Interactive version: theaggregate.ai/model?slug=glm-4-6v · How It Works · Data refreshed daily, snapshot 2026-09-05.