GLM-4.5

Stale
GitHub Python Apache-2.0

Description

GLM-4.5 is the ARC foundation model family from Zhipu AI, combining agentic, reasoning and coding capabilities.

Key Features

  • ARC capabilities — Combines agentic, reasoning and coding in one model
  • Open weights — Full model weights and inference code released
  • Multimodal — Text, image and video input/output support
  • Tool calling — Native function calling for agent framework integration
  • Chinese-leading — Strong performance on Chinese understanding and generation
  • Inference efficient — Multiple quantization and deployment optimizations

Use Cases

💡 Deploying local agent inference services as an open-source base
💡 Providing base model support for Chinese AI agent applications
💡 Serving as backend for coding copilots and agents
💡 Powering vision modules in multimodal agent applications
💡 Replacing closed-source LLMs in private deployment scenarios

Strengths & Limitations

Strengths

  • High community interest (4.4k stars)
  • Permissive open-source license (Apache-2.0)
  • Responsive to issues, low backlog

⚠️ Limitations

  • No updates in over 6 months

Quick Start

# Install inference deps
pip install -r requirements.txt

# Load with transformers
from transformers import AutoModel, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained('THUDM/GLM-4.5')
model = AutoModel.from_pretrained('THUDM/GLM-4.5', trust_remote_code=True)

Related Projects