GLM-4.5V supports a maximum output length of 64K Tokens, a feature that enables it to handle the task of generating extremely long text or complex multimodal content. The long context support capability allows the model to deeply analyze tens of pages of documents, generate complete code files or parse long video content. In order to achieve a balance of efficiency in different scenarios, the model also introduces an innovative 'Thinking Mode' switch, which allows the user to choose between fast response or deep reasoning modes according to the task requirements, with the former being suitable for real-time interactions, and the latter being suitable for scenarios that require complex logical analysis.
This answer comes from the articleGLM-4.5V: A multimodal dialog model capable of understanding images and videos and generating codeThe
English 
简体中文 
日本語 
Deutsch 
Português do Brasil