China’s leading technology companies are expanding their presence in the global artificial intelligence race. ByteDance and Alibaba Cloud have introduced new image generation models aimed at competing with Google’s Nano Banana Pro but for cheap.
Both firms are focusing on improving performance while reducing costs for businesses and individual users.
Seedream 5.0
ByteDance has launched Seedream 5.0, now available for beta testing on Jimeng in China and on CapCut globally.
According to the company, the model improves reasoning capabilities and better interprets complex prompts. It allows users to edit specific elements of an image without regenerating the entire design.
In one demonstration, a snowy night scene was created and later modified by switching lights on and off while preserving the rest of the composition.
The release follows ByteDance’s recent introduction of Seedance 2.0, an AI video model that has been creating Hollywood-grade hyper-realistic videos.
Qwen-Image-2.0
Alibaba Cloud has introduced Qwen-Image-2.0, developed by its Qwen team.
The model combines image generation and editing within a single system. It supports prompts of up to 1,000 tokens and produces 2K resolution images.
Qwen-Image-2.0 is designed to handle structured layouts, multi-panel designs, and consistent characters across scenes. It also demonstrates strong performance in rendering Chinese text and complex calligraphy.
Feature Comparison
Below is a comparison of key features among the three competing models:
| Feature | ByteDance Seedream 5.0 | Alibaba Qwen-Image-2.0 | Google Nano Banana Pro |
|---|---|---|---|
| Developer | ByteDance | Alibaba Cloud (Qwen team) | Google DeepMind (Gemini) |
| Core Function | Text-to-image generation and editing | Unified image generation and editing | Image generation and advanced editing |
| Native Output Resolution | Supports 2K and 4K outputs | Native 2K (2048×2048) output | Up to 4K resolution output |
| Prompt Handling | Designed for detailed prompt understanding | Supports long prompts up to around 1,000 tokens | Advanced prompt-based generation (no official token limit stated) |
| Text Rendering | Generates legible text within images | Strong typography and structured text rendering | Advanced multilingual text rendering |
| Generation and Editing Integration | Supports selective image edits | Generation and editing are integrated in one model | Integrated image creation and editing tools |
| Availability | Beta testing via Jimeng (China) and CapCut (global) | Available via Qwen platforms | Available through Gemini apps and Google AI tools |
| Model Base | Proprietary Seedream model | Qwen multimodal architecture | Built on Gemini 3 Pro Image |
Disclaimer: Feature details are based on publicly available information and may change over time.
