Qwen3.8 Omni Flash 於 2026 年 9 月登上 Qwen AI 平台,可在同一工作流程原生接收文字、圖片、音訊和影片,支援最多 100 萬 token 上下文。[2][1]
阿里巴巴稱,模型在 29 項評測的平均得分較 Qwen3.5 Omni Plus 高逾 25%;長影片亦可按需要查看相關片段,而非持續處理每一格畫面。[12][16]
擴充後的 Qwen MM Plugins、開源 Qwen Live Harness 和 Realtime 版本分別支援長時間及即時互動流程;官方稱每小時音訊及音畫輸入的 API 費用分別下降逾 98% 和逾 93%。[52][1][12]
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
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