GLM-4.5-MoE-106B-A12B-0715
GLM-4.5-MoE-106B-A12B-0715 is an inference model developed by Zhipu AI. It is provided as a foundational model with a 128K context window and a large parameter scale.
Parameters
1060.0B
Context Window
128K
License
Apache 2.0
Release Date
2025-07-28
API Pricing
API pricing for this model is not yet available
Strengths
- ・Advanced reasoning capabilities.
- ・Maintains 128K long context.
- ・Available under an open license.
Weaknesses
- ・Enormous model size.
- ・Requires high computational resources.
- ・Increased operational costs.
Use Cases
- ・Complex logical reasoning tasks.
- ・Analysis of long documents.
- ・Advanced knowledge extraction.
Deep Analysis
Architecture
MoE (106B total, 12B active)
Lightweight GLM-4.5 variant
Context Window
128K tokens
Release Date
July 2025
License
Open weights
Apache 2.0
Specialization
Agent, Reasoning, Coding (ARC)
Purpose-built for agentic apps
Strengths
- ・Efficient 12B active parameters from 106B total
- ・128K context window
- ・Optimized for tool invocation and agents
- ・Supports hybrid reasoning modes (Thinking/Non-Thinking)
- ・Free Flash tier available
- ・Strong coding and agent capabilities
Weaknesses
- ・Less capable than full GLM-4.5 355B
- ・Chinese model with some English limitations
- ・Competing in crowded open-weight space
Competitor Comparison
| Model | Arena | SWE | GPQA | Price |
|---|---|---|---|---|
| GLM-4.5 355B | - | - | - | Higher |
| DeepSeek V3 | - | - | - | Comparable |
| Qwen 2.5 72B | - | - | - | Comparable |
GLM-4.5-Air (106B-A12B) is Zhipu AI's cost-effective model in the GLM-4.5 series. With 106B total and 12B active parameters, it delivers strong reasoning, coding, and agent capabilities at lower cost than the full 355B variant.
Analysis generated: 2026-05-24