ZHIPU AI OPEN PLATFORM

Zhipu·Large Model Platform

We don’t just build models. We build decisions.

GLM-4.5

Zhipu's New Generation Flagship Model. It achieves SOTA in open source models!

GLM-4.5V

Zhipu's most powerful VLM, reaching SOTA for open-source models, covering core tasks such as video understanding, and GUI.

GLM-4.5-Flash

GLM-4.5-Flash is the free version of GLM-4.5, excels at reasoning, coding, agent tasks.

View details

Zhipu Model Family

Model Matrix

View All Models

GLM-4.5 · Flagship Model
Zhipu's new flagship model achieves SOTA among global open-source models with multiple capabilities!
GLM-Z1 · Reasoning Model
Enhanced capabilities in solving a wide range of complex problems and mathematical reasoning.
GLM-4.5V · Vision Model
The world's 100B-Scale open-source visual reasoning model SOTA, covering UI coding, GUI Agent tasks, etc.
GLM-4-Voice · Speech Model
Understand and generate Chinese and English voices directly for real-time voice conversations.
CogView · Image Model
Supports text-to-image generation with comprehensive improvements.
CogVideoX · Video Model
The video quality is clearer and more stable, supports large-scale movement

Live Video Calling

Universal Translation
Fluent in multiple languages, with deep understanding of context, tone, slang, and terminology.
View Details
AI Drawing
Generate images from text descriptions for various uses like platform design, content creation, and event promotion.
View Details
Special Effects Videos
Video templates based on popular gameplay for short video creative production.
View Details
Job Matching
Analyzes resume-job fit from multiple dimensions, accurately identifying key information.
View Details
Script Quality Inspection
Analyze customer service conversations, extract key info, conduct quality checks, and improve service quality.
View Details
Homework Grading
Combining with K12 domain knowledge to provide highly accurate answer evaluation and essay grading capabilities.
View Details

Make Models Understand Your Task Better

Data Efficiency：Upload your dataset with one click.
Flexible Methods：LoRA fine-tuning or full-parameter fine-tuning.
User-Friendly：Low-code framework, training can be completed in as little as 10 minutes.