{"rewrite":{"id":"r_786846900c24a21a72ff8669","clusterId":"c_574078cc1e82589a6481c478","slug":"alibaba-adds-qwen3-8-omni-flash-to-qwen-model-line","model":"deepseek-v4-1-flash","headline":"Alibaba Adds Qwen3.8-Omni-Flash To Qwen Model Line","summary":"Alibaba released Qwen3.8-Omni-Flash, a native omni-modal model in its Qwen line. It supports a 1-million-token context window and keeps text performance equal to text-only models of the same size. Across 29 benchmarks, its average score rose more than 25% over Qwen3.5-Omni-Plus, while API pricing for audio input fell more than 98% and audio-video input more than 93%.","whyItMatters":"Alibaba is pricing audio and video input down hard while closing the benchmark gap to Gemini 3.8 Flash, which puts the cost of agent workflows built on omni-modal models under pressure.","webCardHtml":"\u003cp\u003eThe model runs on Qianwen AI and ships with speech recognition for 74 languages and speech generation for 29. Its reasoning scores include 82.7 on LongAudioSpan, 63.4 on OmniVideoBench, 28.2 on OmniCap-IF, and 89.7 on AliMeeting. On agent, coding, and long-horizon tasks it scored 71.0 on WildClawBench-MM, up 36.5 over Qwen3.5-Omni-Plus, and 69.6 on UniClawBench. Alibaba says it surpassed Gemini 3.8 Flash on multiple tests.\u003c/p\u003e","blueskyPost":"Alibaba cut Qwen3.8-Omni-Flash audio API pricing by more than 98%. The model is a multimodal system with a 1-million-token context window, and the pricing move signals Alibaba is willing to absorb compute costs to win developers.","twitterPost":"Qwen3.8-Omni-Flash matches text-only models of its size while handling audio and video natively. Alibaba's pricing cut puts the squeeze on rivals.","threadsPost":"Alibaba's Qwen3.8-Omni-Flash keeps text performance level with text-only models of the same size, across 29 benchmarks. The audio input API price fell more than 98%. Keeping a multimodal model that cheap means Alibaba is buying developer adoption rather than selling inference margin.","newsletterBlurb":"Alibaba released Qwen3.8-Omni-Flash, a native omni-modal model with a 1-million-token context window. It scores above Qwen3.5-Omni-Plus across 29 benchmarks on average, and audio input API pricing drops more than 98% by comparison. Alibaba says audio and video performance is close to Gemini 3.8 Flash, with overall audio performance exceeding it.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260918-qwen-3-8-omni-flash/\",\"title\":\"Alibaba Releases \\\"Qwen3.8-Omni-Flash,\\\" Achieving Audio and Video Processing Performance Comparable to Gemini 3.8 Flash\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4743,"outputTokens":678,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1789712964,"createdAt":"2026-09-18T06:24:23.000Z","publishedAt":"2026-09-18T06:28:22.000Z","updatedAt":"2026-09-18T06:28:22.000Z"},"cluster":{"id":"c_574078cc1e82589a6481c478","canonicalTitle":"「Qwen3.8-Omni-Flash」リリース、Gemini 3.8 Flashに匹敵する音声・映像処理性能を達成","representativeArticleId":"a_514ad3da26fd7816b94d5d5b","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[\"Qwen3.8-Omni-Flash\"],\"studios\":[\"Alibaba\"],\"people\":[],\"type\":\"release\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-09-18T05:20:00.000Z","lastSeenAt":"2026-09-18T05:20:00.000Z","updatedAt":"2026-09-18T06:28:20.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20260918-qwen-3-8-omni-flash/","title":"「Qwen3.8-Omni-Flash」リリース、Gemini 3.8 Flashに匹敵する音声・映像処理性能を達成"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":["Qwen3.8-Omni-Flash"],"studios":["Alibaba"],"people":[],"type":"release","domain":"other","is_roundup":false},"keyFacts":null}