GLM-4.6

Multimodal
Zero-eval
#1LiveCodeBench v6
#2HLE

by Zhipu AI

+
+
+
+
About

GLM-4.6 was introduced as an enhanced iteration of the GLM-4 series, designed to provide improved capabilities in bilingual language understanding and generation. Built to incorporate refinements to the GLM architecture, it represents continued advancement in Zhipu AI's model development.

+
+
+
+
Pricing Range
Input (per 1M)$0.60 -$0.60
Output (per 1M)$2.00 -$2.00
Providers2
+
+
+
+
Timeline
AnnouncedSep 30, 2025
ReleasedSep 30, 2025
+
+
+
+
Specifications
Capabilities
Multimodal
+
+
+
+
License & Family
License
MIT
Performance Overview
Performance metrics and category breakdown

Overall Performance

7 benchmarks
Average Score
61.2%
Best Score
93.9%
High Performers (80%+)
3

Performance Metrics

Max Context Window
196.6K
Avg Throughput
85.0 tok/s
Avg Latency
1ms
+
+
+
+
All Benchmark Results for GLM-4.6
Complete list of benchmark scores with detailed information
AIME 2025
text
0.94
93.9%
Self-reported
LiveCodeBench v6
text
0.83
82.8%
Self-reported
GPQA
text
0.81
81.0%
Self-reported
SWE-Bench Verified
text
0.68
68.0%
Self-reported
BrowseComp
text
0.45
45.1%
Self-reported
Terminal-Bench
text
0.41
40.5%
Self-reported
HLE
multimodal
0.17
17.2%
Self-reported
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+