[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-推理模型":3},{"tag":4,"articles":9},{"id":5,"name":6,"slug":6,"article_count":7,"description_zh":8,"description_en":8},"83ddd36b-dc43-40bd-9d31-48ef0fe9d77a","推理模型",2,null,[10,19],{"id":11,"slug":12,"title":13,"summary":14,"category":15,"image_url":16,"cover_image":16,"language":17,"created_at":18},"5f593215-e1e5-4ea1-92f8-0a08d0ab97a8","aime-2026-leaderboard-qwen-leads-math-tests-zh","AIME 2026 排行榜：Qwen 先拿下數學測試","AIME 2026 排行榜只有 8 個模型，但分數差很大。Qwen3.6 Plus 以 0.953 領先，最低只有 0.375。這份數學基準很適合看 LLM 的推理穩定度。","research","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775179306019-0ww7.png","zh","2026-04-03T01:21:30.728823+00:00",{"id":20,"slug":21,"title":22,"summary":23,"category":15,"image_url":24,"cover_image":24,"language":17,"created_at":25},"ffa8459f-678e-40b9-a513-dee6b02800bc","arc-prize-leaderboard-cost-performance-zh","ARC 排行榜把成本攤開來看","ARC Prize 排行榜把成本和分數放在同一張圖上，ARC-AGI-3 也把任務拉進互動環境。這篇看它怎麼逼 AI 團隊正視算力、Token 和實際可部署性。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775143855363-h1lx.png","2026-04-02T15:30:39.292235+00:00"]