[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-ai-harness-longharness-lifts-long-document-accuracy-to-60-percent":10,"sections":40},{"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":30,"tags":31,"sources":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},11275,"ai-harness-longharness-lifts-long-document-accuracy-to-60-percent","AI Harness LongHarness Lifts Long Document Accuracy to 60 Percent","LongHarness, a new AI harness, pushes one model's accuracy on brutal long-document tests from single digits to over 60 percent, beating Claude Code and Codex.","A new agent harness called LongHarness claims to fix a basic flaw in how AI agents read huge document piles: they forget what they found.\n\nResearchers built LongHarness as a model-agnostic harness for AI agents reasoning over document corpora exceeding 100 million tokens, far more than any model can hold in context at once. Its two components, Evidence-Conclusion Coevolution and Evidence Lifecycle Management, track what evidence has been found and what conclusions it supports, instead of letting that information get buried in a sprawling interaction history. Tested on benchmarks including OfficeQA Pro V1 and V2, AA-LCR, and LongBench v2, the harness took GLM-5.3-Flash's exact-match accuracy on OfficeQA Pro from 15.8% to 60.2% on V1 and from 4.4% to 61.1% on V2, compared with the same backbone reasoning alone. It also beat Claude Code by 12.0 percentage points on V1 and Codex by 18.9 points on V2, and with a different backbone, Qwen3.8-Flash, reached 61.7% on V1 - 8.3 points ahead of Claude Code and 12.0 ahead of Codex.\n\nThe gains swing wildly across tests - roughly 4x on one OfficeQA Pro version, nearly 14x on another - which says more about how punishing these benchmarks are for agents with no memory discipline than about any single multiplier. The real finding here isn't the percentage, it's the diagnosis: agents don't struggle with long documents because they can't retrieve facts, they struggle because they lose track of what they already found and why it mattered.\n\nCall it a filing system for AI memory - unglamorous, but apparently the part everyone else forgot to build.","[\"ai\",\"ai-agents\",\"long-context\",\"benchmarks\"]","2026-10-12T04:00:00.000Z","2026-10-12T05:21:53.934Z","2026-10-12T05:21:58.820Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"publisher-r1","publisher",1,"The dek claims the accuracy gain 'quadruples,' but the body reports two different baseline\u002Fresult pairs (15.8%→60.2% is ~4x, while 4.4%→61.1% is ~14x), making the dek's single figure inconsistent with the data in the body.","resolved","ai",[30,32,33,34],"ai-agents","long-context","benchmarks",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.13075",0,{"sections":41},[42,45,49,54,59,64,69,74,79,84,89,94,99,104],{"name":43,"slug":30,"count":44,"latest_published_at":18},"AI",6896,{"name":46,"slug":47,"count":48,"latest_published_at":18},"Security","security",946,{"name":50,"slug":51,"count":52,"latest_published_at":53},"Policy","policy",496,"2026-10-11T20:06:43.000Z",{"name":55,"slug":56,"count":57,"latest_published_at":58},"Deals","deals",484,"2026-10-09T12:30:00.000Z",{"name":60,"slug":61,"count":62,"latest_published_at":63},"Hardware","hardware",234,"2026-10-11T17:20:00.000Z",{"name":65,"slug":66,"count":67,"latest_published_at":68},"Science","science",196,"2026-10-11T15:34:40.000Z",{"name":70,"slug":71,"count":72,"latest_published_at":73},"Consumer Tech","consumer-tech",190,"2026-10-11T21:14:40.000Z",{"name":75,"slug":76,"count":77,"latest_published_at":78},"Startups","startups",120,"2026-10-09T17:02:31.000Z",{"name":80,"slug":81,"count":82,"latest_published_at":83},"Software","software",115,"2026-10-11T07:34:45.000Z",{"name":85,"slug":86,"count":87,"latest_published_at":88},"Dev Tools","dev-tools",110,"2026-10-09T13:03:48.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"General","general",70,"2026-10-11T14:15:00.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Gaming","gaming",59,"2026-10-09T11:43:43.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":105,"slug":106,"count":107,"latest_published_at":108},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]