- GLM-5V-Turbo is a multimodal coding foundation model designed for visual programming, aiming to give AI agents "eyes."
- The tech release follows Zhipu's strong 2025 financial performance, where surging demand for AI agents drove a massive 131.9% jump in full-year revenue.
Chinese AI startup Zhipu has released GLM-5V-Turbo, a multimodal coding foundation model built specifically for visual programming.
The model is designed to give AI agents "eyes," further consolidating Zhipu's market position in the enterprise-level agentic engineering sector.
GLM-5V-Turbo integrates visual and text capabilities, breaking the limitations of traditional pure-text programming inputs, according to a Zhipu announcement on Thursday.
It can understand design mockups, web interfaces, and videos, and autonomously generate complete, runnable frontend code projects based on them, Zhipu said.
The new model is adapted for mainstream frameworks such as Claude Code and AutoClaw, extending the agents' perception pipeline from text to visual interaction.
After integrating the model, agents like OpenClaw gain visual perception capabilities, allowing them to autonomously browse web pages and interpret complex candlestick charts.
This technological breakthrough comes as Zhipu sees widening losses due to R&D spending, yet exponential growth in its core business, with total revenue in 2025 reaching 724 million yuan ($105 million), up 131.9% year-on-year.
Internet giants like ByteDance and Meituan have already highly praised GLM-5V-Turbo during internal testing, validating the model's comprehensive capabilities, Zhipu noted.
As desktop agents accelerate their entry into corporate environments, large model technology is transitioning from a mere efficiency-boosting tool into a core digital workforce for businesses.
The company released the GLM-5-Turbo foundation model optimized specifically for OpenClaw last month, addressing the core pain point of general large models easily stalling in complex, long-chain tasks.
Zhipu is attempting to secure its foothold in the fiercely competitive AI market through continuous technological iteration and ecosystem building.
