Alibaba Introduces New Qwen AI Model
阿里巴巴推出全新通義千問 AI 模型
更新於: 2026年8月9日 上午04:15
Alibaba has officially unveiled its latest artificial intelligence powerhouse, Qwen3.8-Max, marking a major milestone in the company's AI development.
阿里巴巴正式發表了其最新的人工智慧中樞系統 Qwen3.8-Max，標誌著該公司在 AI 發展上的一個重要里程碑。
Released in early August 2026, this flagship multimodal model features a massive 2.4-trillion-parameter architecture.
該旗艦級多模態模型於 2026 年 8 月初發布，採用了規模龐大的 2.4 兆參數架構。
By using a 'sparse mixture-of-experts' design, it activates only 95 billion parameters at a time, allowing for faster response times and lower operational costs.
透過「稀疏專家混合」設計，它每次僅激活 950 億個參數，從而實現了更快的響應速度與更低的運營成本。
Designed to handle complex, long-horizon tasks, the model supports an impressive 1-million-token context window, making it highly capable for intricate coding and multi-step workflows.
此模型旨在處理複雜且跨度長遠的任務，支援高達 100 萬個代幣(token)的上下文窗口，使其在處理繁複編碼與多步驟工作流程上表現卓越。
Alibaba has positioned Qwen3.8-Max as a top-tier competitor on the global stage, earning it a high ranking on the Arena.AI platform.
阿里巴巴將 Qwen3.8-Max 定位為全球舞台上的頂級競爭者，並使其在 Arena.AI 平台上獲得了高排名。
Perhaps most exciting for the developer community is Alibaba's decision to release the model weights publicly, returning to an open-source strategy despite testing new, structured monetization models for large commercial users.
對於開發者社群而言，最令人振奮的或許是阿里巴巴決定公開發布模型權重，儘管他們正為大型商業用戶測試新的結構化貨幣化模式，此舉象徵其重返開源策略。
Alongside this release, Alibaba introduced QwenWork, an integrated workplace agent platform designed to streamline enterprise tasks.
隨此發布，阿里巴巴還推出了 QwenWork，這是一個旨在簡化企業任務的整合式工作場所代理平台。
This launch underscores Alibaba’s ambition to bridge the technological gap with Western competitors and solidify the Qwen series as a central pillar in the international AI ecosystem.
此次發布凸顯了阿里巴巴縮小與西方競爭對手技術差距的雄心，並鞏固了 Qwen 系列作為國際 AI 生態系統核心支柱的地位。
