[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-benchmark-tests-how-well-ai-judges-video-understanding":10,"sections":34},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},9293,"new-benchmark-tests-how-well-ai-judges-video-understanding","New Benchmark Tests How Well AI Judges Video Understanding","A new benchmark and two reward models aim to fix the weak, data-starved state of AI video-understanding evaluation.","A team of AI researchers just gave video-understanding reward models their first real yardstick.\n\nThe researchers built VURB, a benchmark of 2,100 preference pairs covering general, long-form, and reasoning-heavy video tasks, each pair annotated with long chain-of-thought reasoning traces averaging 1,143 tokens. They also built VUP-35K, a roughly 35,000-example preference dataset generated through a fully automated pipeline, to train better reward models. Using that data, they trained two reward models: a discriminative one called VideoDRM and a generative one called VideoGRM. Both beat existing approaches on VURB and on a separate test, VideoRewardBench, and both showed bigger gains when given more compute to sample multiple answers before picking the best one.\n\nReward models are the judges that tell other AI systems whether an output is good or bad, and in video that judgment has mostly been guesswork, thanks to a shortage of solid benchmarks and training data. Text and image reward models are mature enough to meaningfully steer model behavior at scale; video has lagged because clips are longer, harder to label, and pricier to evaluate. This work hands video AI the kind of preference data and evaluation standard text models have had for a while.\n\nIt is still a research paper, not a shipped product, so the real test is whether other labs adopt VURB rather than quietly building their own benchmark, which is the usual move in AI research.","[\"ai\",\"reward-models\",\"video-understanding\",\"benchmarks\"]","2026-10-01T04:00:00.000Z","2026-10-02T07:24:35.727Z","2026-10-02T07:24:38.092Z","published",null,[],"ai",[24,26,27,28],"reward-models","video-understanding","benchmarks",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.07872",0,{"sections":35},[36,39,43,47,52,57,61,66,71,75,80,85,90,95],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",5671,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",820,{"name":44,"slug":45,"count":46,"latest_published_at":18},"Policy","policy",430,{"name":48,"slug":49,"count":50,"latest_published_at":51},"Deals","deals",298,"2026-09-30T21:00:26.000Z",{"name":53,"slug":54,"count":55,"latest_published_at":56},"Hardware","hardware",196,"2026-09-30T13:00:00.000Z",{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",165,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",149,"2026-09-30T22:57:11.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Dev Tools","dev-tools",93,"2026-10-01T02:30:48.000Z",{"name":72,"slug":73,"count":69,"latest_published_at":74},"Software","software","2026-09-30T21:41:11.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Startups","startups",84,"2026-09-30T20:39:09.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Gaming","gaming",51,"2026-09-30T16:24:30.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]