Alibaba’s Qwen3 Tops Open-Source AI Rankings
阿里巴巴 Qwen3 登顶开源 AI 模型榜首
Alibaba’s Qwen3 has overtaken DeepSeek’s R1 as the top open-source AI model, according to LiveBench benchmark tests. These tests assess coding, math, and data analysis capabilities. Released by Alibaba Cloud, the Qwen3 family includes eight models ranging from 600 million to 235 billion parameters. The achievement highlights Alibaba’s rapid progress in AI and its rising global influence in open-source innovation. While Qwen3 leads among open models, it still trails behind closed-source giants like OpenAI’s o3, Google’s Gemini Pro 2.5, and Anthropic’s Claude 3.7.
阿里巴巴的 Qwen3 模型在 LiveBench 基准测试中超越 DeepSeek 的 R1，成为全球排名最高的开源 AI 模型。该测试评估了模型在编程、数学和数据分析等方面的能力。Qwen3 由阿里云发布，共包括八个模型，参数规模从 6 亿到 2350 亿不等。这一成就展现了阿里在 AI 领域的飞速发展，以及其在全球开源社区中日益增强的影响力。尽管 Qwen3 在开源模型中排名第一，但仍落后于 OpenAI 的 o3、谷歌的 Gemini Pro 2.5 和 Anthropic 的 Claude 3.7 等闭源模型。
