China just unveiled an AI model that could upend America’s tech advantage

Chinese AI startup DeepSeek launched V4-Flash on Friday, the cheapest major AI model to operate at an average cost of 3 cents per test, according to research firm Artificial Analysis, which found it costs more than 100 times less to run than Anthropic's Claude Fable 5. The launch intensifies pressure on U.S. AI companies like OpenAI and Anthropic, as Chinese rivals including Alibaba, Moonshot AI, MiniMax, and Z.AI compete on cost and efficiency. Alibaba also unveiled its largest AI model, Qwen3.8-Max, on Monday, capable of processing up to 1 million tokens at once.

China just unveiled an AI model that could upend America’s tech advantage DeepSeek is reportedly looking to regain momentum after being surpassed by Chinese rivals - Bookmark - CommentsGo to comments Chinese AI /topic/ai startup DeepSeek /tech/artificial-intelligence-china-deepseek-chatgpt-openai-b3025304.html has unveiled a new model it says could shake up the global artificial intelligence race by dramatically undercutting rivals on cost. DeepSeek's new V4-Flash model, launched Friday, is the cheapest major AI model to operate, according to research firm Artificial Analysis. The firm found it costs more than 100 times less to run than Anthropic's Claude Fable 5, renewing pressure on U.S. AI companies spending billions to stay ahead. The launch comes amid increasing concerns that Chinese companies are developing powerful artificial intelligence systems that run vastly more quickly and cheaply than those made by their US competitors. Artificial Analysis estimated V4-Flash costs an average of just 3 cents per test, compared with 86 cents for Chinese rival Moonshot AI's Kimi K3, $1.86 for OpenAI's GPT-5.6 Sol and $3.15 for Claude Fable 5 /news/world/americas/us-politics/anthropic-mythos-5-fable-offline-ai-trump-order-b2995199.html . The comparison factors in the total amount of data a model processes to complete a task, offering a broader measure of real-world operating costs. The launch comes as DeepSeek reportedly prepares for a potential initial public offering. Its earlier R1 model rattled global markets in early 2025 by showing that powerful AI systems could be built far more cheaply than many analysts expected. While V4-Flash is the lowest-cost model tested, it is not the highest-performing. It scored 50 out of 100 on Artificial Analysis' Intelligence Index, matching Google's Gemini /tech/google-gemini-robot-deepmind-ai-b3025138.html 3.6 Flash and trailing Moonshot's Kimi K3 and leading models from OpenAI and Anthropic /news/openai-anthropic-chatgpt-san-francisco-b3025090.html . DeepSeek is also developing a more powerful V4-Pro model. The release highlights the intensifying AI battle between China and the United States. Chinese firms including Alibaba, Moonshot AI, MiniMax and Z.AI are racing to produce cheaper, more efficient models that can compete globally with OpenAI, Anthropic and Google. Alibaba added to that competition on Monday by unveiling its largest AI model yet, Qwen3.8-Max. The multimodal system can process text, images and video and is designed to handle up to 1 million tokens at once. Alibaba said the model uses a "mixture-of-experts" architecture to reduce computing costs while maintaining performance. The latest launches underscore how China's AI developers are increasingly competing not only on capability but on price, intensifying pressure on Western rivals as businesses look for lower-cost ways to deploy artificial intelligence at scale. Join our commenting forum Join thought-provoking conversations, follow other Independent readers and see their replies Comments comments-area