Qwen3.8 Omni Flash 於 2026 年 9 月在 Qwen AI 平台推出,可在同一流程接收文字、圖片、音訊與影片,上下文視窗最高達 100 萬 token。[1][2]
阿里巴巴稱,相較 Qwen3.5 Omni Plus,新模型在 29 項評測的平均分數提升逾 25%;長影片可按任務需要選取片段檢視,而非一律處理每一格畫面。[12][16]
延伸的 Qwen MM Plugins、開源的 Qwen Live Harness 與獨立的 Realtime 版本,分別支援長流程及即時互動;官方稱每小時音訊、音影片輸入 API 成本分別下降逾 98% 和 93%。[1][12][52]
What is Alibaba’s Qwen3.8 Omni Flash, launched on the Qwen AI platform in September 2026, and how do its native text, image, audio, and videAI-generated editorial hero image for What is Alibaba’s Qwen3.8 Omni Flash, launched on the Qwen AI platform in September 2026, and how do its native text, image, audio, and vide.
AI 提示詞
Create a landscape editorial hero image for this Studio Global article: What is Alibaba’s Qwen3.8 Omni Flash, launched on the Qwen AI platform in September 2026, and how do its native text, image, audio, and vide. Article summary: Alibaba’s Qwen3.8 Omni Flash is a September 2026 model available on the Qwen AI platform that accepts text, images, audio, and video natively in one workflow, with a context window of up to 1 million tokens.. Topic tags: general web, agents, ai, workflow, productivity. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake