[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-tool-use":3},{"tag":4,"articles":11},{"id":5,"name":6,"slug":7,"article_count":8,"description_zh":9,"description_en":10},"95d84e59-4a34-4d99-a04c-cf3218a596d4","tool use","tool-use",3,"工具使用指的是讓 LLM 或代理在需要時呼叫搜尋、程式執行、資料庫或外部 API，而不只是靠模型內部生成。重點不在多會用工具，而在何時該用、何時該停，牽涉推理正確率、延遲、成本與可觀測性，尤其在多模態代理與產品化工作流中更明顯。","Tool use refers to letting LLMs and agents call search, code execution, databases, or external APIs only when needed, instead of relying on generation alone. The hard part is deciding when a tool helps and when it just adds latency, cost, or failure modes, especially in multimodal agents and production workflows.",[12,21],{"id":13,"slug":14,"title":15,"summary":16,"category":17,"image_url":18,"cover_image":18,"language":19,"created_at":20},"f54893ef-5e99-4326-9120-88752dbc8b4d","why-kimi-k25-changes-open-source-agent-race-en","Why Kimi K2.5 Changes the Open-Source Agent Race","Kimi K2.5 makes open-source agents matter by pairing multimodal reasoning with tool-heavy execution.","model-release","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1777965052357-hsby.png","en","2026-05-05T07:10:25.728264+00:00",{"id":22,"slug":23,"title":24,"summary":25,"category":26,"image_url":27,"cover_image":27,"language":19,"created_at":28},"3cefc37f-e116-4597-a5cb-55bfb3fc4aa4","act-wisely-tool-use-agentic-multimodal-models-en","Act Wisely: Teaching Agents When Not to Call Tools","A new training scheme, HDPO, aims to cut blind tool use in multimodal agents by separating accuracy from tool efficiency.","research","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775801032138-7jih.png","2026-04-10T06:03:34.728615+00:00"]