ByteDance takes on Google’s Nano Banana as China pushes back in the AI image race

ByteDance

• ByteDance launches Seedream 5.0 to compete with Google's Nano Banana Pro in image editing.
• Seedream 5.0 offers stronger reasoning and precise edits at a lower price
• Alibaba Cloud launched Qwen-Image-2.0, enhancing China's generative AI race.

Did our AI summary help?

ByteDance is escalating China’s challenge to US-led generative AI tools with the launch of Seedream 5.0, a new image-generation and editing model designed to undercut Google’s Nano Banana on both price and capability.

The Chinese short-video giant said Seedream 5.0 offers stronger reasoning, higher accuracy, and more precise edit controls, features that have helped Nano Banana Pro gain global traction. In a post on CapCut’s official X account, ByteDance framed the comparison bluntly: “Think Nano Banana Pro, but cheaper.”

Seedream 5.0 is now available for beta testing on Jimeng and CapCut, ByteDance’s domestic and international AI content creation platforms. The company says the model is better at understanding nuanced text prompts and allows users to retouch specific parts of an image without regenerating the entire scene, a workflow advantage that has become table stakes in high-end image editors.

In a test conducted by the South China Morning Post, Seedream 5.0 successfully interpreted a prompt to generate “a snowy night with a sense of quietude,” producing a subdued scene featuring a snow-covered house and dim exterior lighting. A follow-up prompt to switch off the outdoor lamp and light the interior instead was executed accurately, suggesting improved contextual reasoning across multiple edits.

ByteDance’s announcement landed on the same day Alibaba Cloud released Qwen-Image-2.0, its latest image model and the first in the Qwen line to combine image generation and editing in a single system. Alibaba said the model uses a lightweight architecture for faster output and performs better than Google’s Nano Banana Pro when rendering Chinese characters, a long-standing weakness for Western models.

Together, the launches underscore how quickly Chinese tech firms are iterating in response to US advances in generative AI. While American companies have dominated mindshare, Chinese players are increasingly competing on cost, localisation, and integration with existing creator platforms.

Seedream 5.0 follows the debut of ByteDance’s Seedance 2.0 video model over the weekend, which drew attention for producing highly realistic video content. Analysts say these tools could significantly reduce production costs in filmmaking, advertising, and social media content creation.

Both ByteDance and Alibaba position their new image models as tools for individuals and businesses alike, lowering the barrier to professional-grade visuals. In practical terms, that makes them appealing not just to creators, but to small firms and marketers priced out of premium US tools.