GLM-4.6V

GLM-4.6V is the latest open-source multimodal model from GLM (Zhipu AI), featuring a 128K context window that allows processing of large documents, images, and video frames in a single pass. It extends GLM-4V with native function calling, enabling the model to not only understand visual inputs but also invoke tools, APIs, and code executors to perform actions autonomously. This makes it particularly suitable for complex agentic tasks such as web search, data extraction, and coding. The model is available under a permissive open-source license (Apache 2.0) on GitHub and Hugging Face, with commercial use allowed. GLM-4.6V is designed for developers and researchers building autonomous agents, RAG pipelines, and multimodal applications that require both perception and action. Its unique selling point is the seamless integration of vision-language understanding with executable function calls, bridging the gap between seeing and doing. The model supports multi-turn conversations, image input, and fine-tuning, making it a versatile foundation for advanced use cases. In benchmarks, GLM-4.6V demonstrates competitive performance on multimodal tasks like VQA, captioning, and OCR, while also showing strong results in function-calling scenarios compared to proprietary models.