Chinese e-commerce and cloud provider Alibaba launched Qwen 3.8-Max, a 2.4 trillion-parameter model that competes with top models from Anthropic and OpenAI, intensifying the AI arms race on Monday. The release follows DeepSeek V4 Flash 0731, which independent benchmarks by Artificial Analysis show performs within a single point of OpenAI’s GPT-5.6 Luna while costing 40 percent less per task, and at 284 billion parameters, runs on modest enterprise servers. Chinese developers Moonshot, Alibaba, and DeepSeek are attacking US counterparts on price and performance, as US firms like Anthropic and OpenAI raise concerns over the origins and safety of China-made models. Anthropic CEO Dario Amodei stated he opposes open models made in China, distilled from proprietary models, or lacking rigorous safety metrics. Hugging Face CEO Clément Delangue told CNBC Monday that China is “clearly dominating on open models right now” and could dominate at the frontier by the end of this year or next. America’s most capable open weights model, Inkling, is just shy of a billion parameters and cannot keep up with DeepSeek’s finest, while Moonshot's Kimi K3 and Z.ai's GLM 5.2 are in a different orbit. Alibaba is making its most capable model weights available for download, having previously locked top models behind an API.