[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-method-swaps-chain-of-thought-text-for-looped-latent-reasoning":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},10875,"new-method-swaps-chain-of-thought-text-for-looped-latent-reasoning","New Method Swaps Chain-of-Thought Text for Looped Latent Reasoning","A new technique swaps step-by-step chain-of-thought text for parallel latent refinement, slashing reasoning latency without hurting accuracy.","A new AI reasoning method skips the wall of step-by-step text models write before answering, and does the thinking in parallel instead.\n\nResearchers built a system called LLoCoT that replaces the usual chain-of-thought approach, where a model writes out its reasoning one token at a time, with a looped transformer that refines a compact set of latent, non-text representations all at once. The same transformer layer runs for a few refinement passes over this workspace, updating it based on the prompt, before a probabilistic head samples latent tokens in parallel. Those latent tokens then feed an ordinary autoregressive decoder to produce the final answer. Trained and tested on the HumanEval and MBPP coding benchmarks, LLoCoT matched the accuracy of a model fine-tuned on explicit chain-of-thought reasoning, while beating a plain base model, an answer-only fine-tuned model, and a non-autoregressive baseline.\n\nThe appeal is speed: compared to the explicit chain-of-thought baseline, LLoCoT delivers the first answer token about 36 times faster and cuts reasoning-phase latency by roughly 42 times, while end-to-end throughput rises 9.2%. For anyone who has waited through a model's verbose step-by-step preamble, that is the part that matters: comparable answer quality, far less waiting.\n\nChain-of-thought prompting made models look smarter by making them show their work, but generating that work token by token is slow and expensive. This result is part of a wider push to keep the accuracy gains of extra computation without paying for them in generated text. Whether looped latent reasoning holds up beyond two coding benchmarks, and against other labs chasing the same speedup, is the open question.","[\"ai\",\"chain-of-thought\",\"latent-reasoning\",\"research\"]","2026-10-09T04:00:00.000Z","2026-10-09T19:31:34.186Z","2026-10-09T19:31:38.478Z","published",null,[],"ai",[24,26,27,28],"chain-of-thought","latent-reasoning","research",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.11472",0,{"sections":35},[36,39,43,48,53,58,62,67,72,77,82,87,92,97],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",6619,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",927,{"name":44,"slug":45,"count":46,"latest_published_at":47},"Policy","policy",486,"2026-10-08T22:40:11.000Z",{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",474,"2026-10-08T22:00:00.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":57},"Hardware","hardware",229,"2026-10-08T20:47:10.000Z",{"name":59,"slug":60,"count":61,"latest_published_at":18},"Science","science",192,{"name":63,"slug":64,"count":65,"latest_published_at":66},"Consumer Tech","consumer-tech",181,"2026-10-08T23:26:35.000Z",{"name":68,"slug":69,"count":70,"latest_published_at":71},"Startups","startups",117,"2026-10-08T16:45:00.000Z",{"name":73,"slug":74,"count":75,"latest_published_at":76},"Software","software",114,"2026-10-08T17:57:01.000Z",{"name":78,"slug":79,"count":80,"latest_published_at":81},"Dev Tools","dev-tools",105,"2026-10-07T16:59:11.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":86},"General","general",66,"2026-10-09T04:46:11.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"Gaming","gaming",58,"2026-10-08T20:08:45.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]