Glm Image

What is Glm Image?

GLM Image is a next‑generation AI image model combining a 9B autoregressive core with a 7B DiT diffusion decoder to balance global instruction understanding and local detail. It delivers open‑source SOTA text rendering (Word Accuracy 0.9116, NED 0.9557), supports multiple ratios (1:1, 3:4, 4:3, 16:9) and resolutions from 512 to 2048 px, and excels in commercial posters, PPT graphics, popular science illustrations, and multi‑panel comics. Fast generation, industrial‑grade quality, and comprehensive APIs (Python/Java) make integration simple for teams and products.

Key features

• Hybrid autoregressive + diffusion architecture
• SOTA text rendering (0.9116 Word Accuracy, 0.9557 NED)
• Multi‑resolution support (512–2048 px; 1:1, 3:4, 4:3, 16:9)
• Fast generation with industrial‑grade quality
• Comprehensive APIs and SDKs for easy integration

Use cases

• Commercial posters and promotional graphics
• Popular science illustrations with flowcharts and annotations
• Consistent multi‑panel comics and e‑commerce image sets
• Social media cover designs and complex layouts
• PPT visuals and product showcase images