Chinese AI Firms Lead Video Generation with Short-Video Ecosystems and Lower Costs

According to analysts, Chinese AI firms are surpassing their American counterparts in the realm of AI video creation. This edge is largely due to extensive short-video platforms, more lenient copyright regulations, and competitive pricing, which are deemed more vital than sheer computational power. Currently, Alibaba, MiniMax, and ByteDance lead the benchmark rankings on the industry assessment site Artificial Analysis. Alibaba's Wan 3.0 holds the top spot in the text-to-video category with audio, followed by Google Gemini Omni Flash. MiniMax's post-trained H3 version ranks third, with the original open-weight H3 and ByteDance's Seedance 2.0 trailing. Eight of the top ten models are Chinese, which also excel in image-to-video and video editing, as evidenced by user comparisons. Analysts highlight proprietary training data, swift iteration, and strong demand for short-form content as crucial elements. While US giants like OpenAI and Anthropic dominate in closed-source large language models, Chinese companies are becoming significant players in AI video generation.