{"rewrite":{"id":"r_38f5254d3df5ee9e0be0db43","clusterId":"c_c0391bcb8ba8b99e30e0c817","slug":"nvidia-releases-nemotron-3-5-super-vl-an-open-weight-model-for-text-images-and-video","model":"deepseek-v4-1-flash","headline":"NVIDIA Releases Nemotron 3.5 Super VL, An Open-Weight Model For Text, Images, And Video","summary":"NVIDIA released Nemotron 3.5 Super VL on October 15, 2026, an open-weight model that reads text, images, and video in one system and operates external tools. It carries 120 billion parameters and, through mixture-of-experts routing, uses 12 billion per token. NVIDIA claims up to five times the throughput of open models at the same scale. Scale AI tuned a version for Skydance that cut processing costs to about one-fortieth.","whyItMatters":"The Skydance case is the part worth watching: a tuned open-weight model holding accuracy while cutting processing costs to about one-fortieth reframes what the open-weight tier can be sold for against closed API pricing.","webCardHtml":"\u003cp\u003eThe parameter count is the smaller number. Of the 120 billion parameters in Nemotron 3.5 Super VL, mixture-of-experts routing keeps 12 billion active per token, and NVIDIA reports throughput above 600 tokens per second. On 10,000 software development tasks across four B200 GPUs, the model finished in about 10 hours at roughly 51% accuracy, up to four times faster than other models at the same scale by NVIDIA\u0026#39;s account.\u003c/p\u003e\u003cp\u003eGraph reading landed at 80.8% on CharXiv, rising to 84% when Python is used to analyze the charts, against 63% for NVIDIA\u0026#39;s Nemotron 3 Nano Omni.\u003c/p\u003e","blueskyPost":"Nemotron 3.5 Super VL activates 12 billion of its 120 billion parameters per token, which is why the pricetag story exists. Scale AI's Skydance tune cut costs to one-fortieth without shrinking the model.","twitterPost":"Nemotron 3.5 Super VL carries 120 billion parameters but fires 12 billion per token, so the one-fortieth cost cut is a routing result, not a smaller model.","threadsPost":"Nemotron 3.5 Super VL is a 120 billion parameter model that only runs 12 billion per token. That mixture-of-experts routing is the reason Scale AI's Skydance tune reached a fortieth of the cost. The open-weight label and the efficiency are the same design choice.","newsletterBlurb":"NVIDIA released Nemotron 3.5 Super VL on October 15, an open-weight model that handles text, images, and video in one system and can drive external tools. It has 120 billion parameters with 12 billion active per token. Scale AI tuned a version for Skydance that reportedly matched existing accuracy at about one-fortieth the processing cost.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20261016-nemotron-3-5-super-vl/\",\"title\":\"NVIDIA Releases \\\"Nemotron 3.5 Super VL,\\\" an Open-Weight AI That Understands Images, Video, and Text Together, With a Case Study Cutting Processing Costs to One-Fortieth\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4677,"outputTokens":666,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1792129035,"createdAt":"2026-10-16T05:32:08.000Z","publishedAt":"2026-10-16T05:36:08.000Z","updatedAt":"2026-10-16T05:36:08.000Z"},"cluster":{"id":"c_c0391bcb8ba8b99e30e0c817","canonicalTitle":"NVIDIAが画像・動画・文章をまとめて理解するオープンウェイトAI「Nemotron 3.5 Super VL」を公開、処理コストを40分の1に削減する事例も","representativeArticleId":"a_5b1e8474b5f5fa8400d43d6a","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":false,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[\"NVIDIA\"],\"people\":[],\"type\":\"announcement\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-10-16T04:10:00.000Z","lastSeenAt":"2026-10-16T05:40:00.000Z","updatedAt":"2026-10-16T05:44:25.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20261016-nvidia-nemotron-3-5-bolt-74b-a9b-24gb/","title":"NVIDIAがAIエージェントによるタスクをローカルで実行するための「Nemotron 3.5 Bolt 74B-A9B 24GB」を発表"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":["NVIDIA"],"people":[],"type":"announcement","domain":"other","is_roundup":false},"keyFacts":null}