Using Zhipu AI's online CogView4 model to generate an image, the result met expectations, but unfortunately, the text in the image was entirely in English instead of the specified Chinese.

The open-source version wasn't tested, but the online version should theoretically be more powerful.

It seems like the prompt might have been too complex for it to understand or follow, or perhaps the model still prioritizes English internally?

Zhipu AI Entry https://bigmodel.cn/trialcenter/modeltrial

Prompt Used

Please draw an image:
### Overall Layout
- Simple cartoon style
- The image is divided into two parts: left side labeled "Before OpenAI," right side labeled "After OpenAI," connected by an arrow (→) in the middle.
- Each part contains two scenes (top: coding, bottom: debugging)

### Left Side: Before OpenAI
1. **Top Part: Developer Coding**
   - Background: A simple desk with an old-fashioned computer monitor.
   - Character: A cartoon developer (round head) sitting in front of the computer, with a focused and slightly confused expression.
   - Text: In a speech bubble above the developer's head or at the top of the scene, write: "Development Coding - 2 hours".

2. **Bottom Part: Developer Debugging**
   - Background: Same desk and computer, but the developer looks tired and frustrated, holding their head in their hands, staring at the computer screen.
   - Character: Same cartoon developer, with a pained expression.
   - Text: In a speech bubble above the developer's head or at the top of the scene, write: "Developer Debugging - 6 hours".

### Right Side: After OpenAI
1. **Top Part: ChatGPT Generating Code**
   - Background: Same desk and computer, but with a ChatGPT icon possibly next to the computer screen.
   - Character: Developer sitting in front of the computer, with a relaxed or surprised expression, indicating the code was generated by ChatGPT.
   - Text: In a speech bubble above the developer's head or at the top of the scene, write: "ChatGPT Generates Code - 5 minutes".

2. **Bottom Part: Developer Debugging**
   - Background: Desk and computer, with the developer looking even more tired and desperate, holding their head in their hands.
   - Character: Same cartoon developer, with a more pained expression.
   - Text: In a speech bubble above the developer's head or at the top of the scene, write: "Developer Debugging - 24 hours".

Actual Generated Images

Testing with a Simple Prompt Worked Well

It seems that for complex multi-scene prompts, the text embedding support isn't quite there yet. However, for simple scenes, especially those like beaches or advertisements, it works very well with excellent results.