GLM-4.6 FP8 — benchmark results
GLM-4.6 evaluated in FP8 quantization. Provider: Zhipu. Released 2025-09-30. Access: Open.
Unified ELO 1682 ± 21, rank #305 of 1776 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MATH-MC Level 5 | 99.69 | Accuracy (%) | 90.4 |
| RewardBench 2 Math | 90.44 | Accuracy (%) | 89.2 |
| RewardBench 2 Focus | 90.91 | Accuracy (%) | 81.4 |
| MATH-MC Level 1 | 98.6 | Accuracy (%) | 75 |
| MATH-MC Level 4 | 99.33 | Accuracy (%) | 75 |
| MATH-MC Level 2 | 98.53 | Accuracy (%) | 71.3 |
| RewardBench 2 Factuality | 77.95 | Accuracy (%) | 66.7 |
| RewardBench 2 Safety | 87.11 | Accuracy (%) | 64.7 |
| RewardBench 2 Precise IF | 64.06 | Accuracy (%) | 62.7 |
| GSM-MC | 99.01 | Accuracy (%) | 61.2 |
| MATH-MC Level 3 | 98.65 | Accuracy (%) | 52.9 |
| JudgeBench Coding | 91.67 | Accuracy (%) | 47.1 |
Interactive version: theaggregate.ai/model?slug=glm-4-6-fp8 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.