[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-memory-trick-helps-ai-agents-reread-long-documents":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},11354,"new-memory-trick-helps-ai-agents-reread-long-documents","New Memory Trick Helps AI Agents Reread Long Documents","MemReread lets AI agents reread sections of long documents their compressed memory dropped, lifting accuracy on multi-hop question answering.","A new AI technique lets language models reread parts of a long document instead of guessing from a compressed memory.\n\nResearchers propose MemReread, a method for agents that handle long documents by streaming text chunk by chunk and keeping an updating memory instead of attending to the whole thing at once. When that final memory doesn't have enough to answer a question, the system breaks the question into parts and goes back to reread the specific sections it may have discarded earlier. The team paired this with a reinforcement learning setup that decides how many rereading passes a task needs, so easy questions cost less compute than hard ones. A 4 billion parameter version, MemReread-4B, beat strong baselines on long-context reasoning benchmarks, including a 1.6 percentage point edge over DeepSeek-V4-Pro on 2WikiMultiHopQA, while keeping the linear-time cost that makes streaming memory appealing in the first place.\n\nLong-context tools usually force a trade-off: full attention remembers everything but gets slow and expensive as documents grow, while compressed memory stays fast but quietly drops details an answer might need later. MemReread's bet is that letting a model notice its own memory gaps and selectively reread, rather than retrieve blindly, fixes errors caused by overwriting evidence too early.\n\nIt's a benchmark result, not a shipped product, and the headline comparison is to one rival model on one dataset, so treat \"beats strong baselines\" as an early data point rather than a verdict on long-context reasoning generally.","[\"ai\",\"long-context\",\"llm-research\",\"memory\"]","2026-10-12T04:00:00.000Z","2026-10-12T09:56:42.642Z","2026-10-12T09:56:47.985Z","published",null,[],"ai",[24,26,27,28],"long-context","llm-research","memory",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.10268",0,{"sections":35},[36,40,45,50,55,60,65,70,75,80,85,90,95,100],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",6967,"2026-10-12T15:20:00.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",954,"2026-10-12T14:20:37.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",496,"2026-10-11T20:06:43.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Deals","deals",492,"2026-10-12T14:55:53.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",237,"2026-10-12T13:24:45.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Science","science",196,"2026-10-11T15:34:40.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Consumer Tech","consumer-tech",192,"2026-10-12T14:07:13.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Startups","startups",121,"2026-10-12T13:22:34.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Software","software",115,"2026-10-11T07:34:45.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Dev Tools","dev-tools",111,"2026-10-12T11:22:24.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"General","general",71,"2026-10-12T16:10:40.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"Gaming","gaming",59,"2026-10-09T11:43:43.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"Reviews","reviews",35,"2026-10-12T14:00:00.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]