[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-three-multimodal-models-work-in-claude-code-zh":3,"article-related-three-multimodal-models-work-in-claude-code-zh":33,"series-tools-563c146c-b078-4610-93fa-af399a02c89a":83},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":24,"views":29,"created_at":30,"published_at":31,"topic_cluster_id":32},"563c146c-b078-4610-93fa-af399a02c89a","three-multimodal-models-work-in-claude-code-zh","Claude Code 現在能接三個多模態模型","\u003Cp data-speakable=\"summary\">三個多模態模型現在可透過 \u003Ca href=\"\u002Ftag\u002Fopenai\">OpenAI\u003C\u002Fa> 風格設定接入 \u003Ca href=\"\u002Ftag\u002Fclaude-code\">Claude Code\u003C\u002Fa>，也能搭配其他客戶端使用。\u003C\u002Fp>\u003Cp>講白了，這件事重點不是「又多了模型」。重點是，\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-code\" target=\"_blank\" rel=\"noopener\">Claude Code\u003C\u002Fa> 這類工具，開始把多模態模型接進同一套設定流程。你不用為每個模型重寫一份連線邏輯。\u003C\u002Fp>\u003Cp>對開發者來說，這很實際。圖片、影片、\u003Ca href=\"\u002Ftag\u002Ftoken\">Token\u003C\u002Fa> 一起進來，測試成本會低很多。尤其是你在比對不同模型時，切換方式越\u003Ca href=\"\u002Fnews\u002Fgemini-atlas-physical-ai-update-zh\">接近\u003C\u002Fa>，結果越好整理。\u003C\u002Fp>\u003Cp>這篇文章就看兩件事。第一，這三個模型到底怎麼接。第二，這種 OpenAI-style 設定，為什麼會\u003Ca href=\"\u002Fnews\u002Fred-hat-ai-mavenir-telco-ai-stack-zh\">變成\u003C\u002Fa>很多客戶端的共同語言。\u003C\u002Fp>\u003Ctable>\u003Cthead>\u003Ctr>\u003Cth>項目\u003C\u002Fth>\u003Cth>內容\u003C\u002Fth>\u003Cth>意義\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd>模型數量\u003C\u002Ftd>\u003Ctd>3 個\u003C\u002Ftd>\u003Ctd>可直接做橫向比較\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>輸入型態\u003C\u002Ftd>\u003Ctd>Token、圖片、影片\u003C\u002Ftd>\u003Ctd>同時處理文字與多媒體\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>設定方式\u003C\u002Ftd>\u003Ctd>OpenAI-style\u003C\u002Ftd>\u003Ctd>降低客戶端整合成本\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>工具範圍\u003C\u002Ftd>\u003Ctd>Claude Code、OpenClaw、Hermes\u003C\u002Ftd>\u003Ctd>可跨不同前端使用\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003Ch2>OpenAI-style 設定，才是重點\u003C\u002Fh2>\u003Cp>很多人看到新模型，第一反應是「又來了」。但真正值錢的，是接法一致。只要設定格式統一，開發者就能把不同模型放進同一個工作流。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781892161193-9rla.png\" alt=\"Claude Code 現在能接三個多模態模型\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>這對 CLI 工具特別有感。你今天在 \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FOpenClaw\u002FOpenClaw\" target=\"_blank\" rel=\"noopener\">OpenClaw\u003C\u002Fa> 測，明天換到 \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fpythops\u002Fhermes\" target=\"_blank\" rel=\"noopener\">Hermes\u003C\u002Fa>，不用重學一套怪設定。少掉的不是幾行字，是很多試錯時間。\u003C\u002Fp>\u003Cp>再講直接一點。模型本身會變，但 API 習慣不能太亂。誰先把介面做乾淨，誰就比較容易被拿來當預設工具。\u003C\u002Fp>\u003Cul>\u003Cli>統一設定，減少整合成本\u003C\u002Fli>\u003Cli>同一套流程可測多個模型\u003C\u002Fli>\u003Cli>CLI 與其他客戶端更好共用\u003C\u002Fli>\u003Cli>開發者切換速度更快\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>三個多模態模型，適合拿來比\u003C\u002Fh2>\u003Cp>這次的核心不是單一模型性能，而是三個模型一起進場。只要輸入格式一致，開發者就能直接比圖片理解、影片摘要、文字推理。\u003C\u002Fp>\u003Cp>這種比較方式很有價值。因為多模態模型最怕的不是跑不動，而是評估不公平。你如果每次都換介面，最後很難知道差異來自模型，還是來自前處理。\u003C\u002Fp>\u003Cp>說真的，這也是很多團隊最常踩的坑。表面上是在比模型，實際上是在比誰家的 wrapper 寫得比較順。\u003C\u002Fp>\u003Cblockquote>「The future is already here — it's just not evenly distributed.」— William Gibson\u003C\u002Fblockquote>\u003Cp>這句話放在多模態工具鏈上，很貼切。功能不是沒有，而是分散在不同工具裡。現在把接法拉近，才比較像真的能用。\u003C\u002Fp>\u003Cp>對台灣團隊來說，這代表 PoC 速度會更快。你可以先用同一套設定跑三個模型，再決定要不要上線。這比一開始就綁死某一家舒服很多。\u003C\u002Fp>\u003Ch2>和其他客戶端比，差在整合手感\u003C\u002Fh2>\u003Cp>如果只看功能列表，很多工具都差不多。真正拉開差距的，是整合手感。你要的是少改設定、少補文件、少處理奇怪錯誤。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781892163628-7bxp.png\" alt=\"Claude Code 現在能接三個多模態模型\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-code\" target=\"_blank\" rel=\"noopener\">Claude Code\u003C\u002Fa> 的優勢，在於它本來就很靠近開發者工作流。現在再接上多模態模型，對寫程式、看截圖、讀影片片段，都更順手。\u003C\u002Fp>\u003Cp>但別太快高潮。模型接得上，不代表品質就一定好。你還是得看上下文長度、圖片解析、影片切片策略，還有回傳格式穩不穩。\u003C\u002Fp>\u003Cul>\u003Cli>Claude Code：適合程式開發流程\u003C\u002Fli>\u003Cli>OpenClaw：適合替代型客戶端測試\u003C\u002Fli>\u003Cli>Hermes：適合本地或自訂工作流\u003C\u002Fli>\u003Cli>OpenAI-style：最容易跨工具複用\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>這波其實在比「誰比較像標準」\u003C\u002Fh2>\u003Cp>多模態模型很多。真正少的是標準化接法。當大家都開始支援類似 OpenAI 的設定格式，開發者就會自然往這種規格靠攏。\u003C\u002Fp>\u003Cp>這件事很現實。你不一定會記得某個模型名稱，但你會記得哪種接法最省事。工具生態最後拼的，常常不是模型分數，而是誰比較不麻煩。\u003C\u002Fp>\u003Cp>從產業角度看，這也讓 API 供應商更難只靠封閉格式吃市場。只要客戶端能快速切換，模型差異就會被攤在桌上比。這對使用者是好事。\u003C\u002Fp>\u003Cp>如果你在做產品，現在最值得做的事很簡單。把模型層抽乾淨，把設定寫清楚，把替換成本壓低。這樣你才不會被某一家綁住。\u003C\u002Fp>\u003Ch2>接下來，先看兩個方向\u003C\u002Fh2>\u003Cp>第一個方向，是實際 \u003Ca href=\"\u002Ftag\u002Fbenchmark\">benchmark\u003C\u002Fa>。不要只看宣傳詞，直接比圖片理解、影片摘要、程式碼修正。這些才是開發者每天會碰到的場景。\u003C\u002Fp>\u003Cp>第二個方向，是客戶端支援度。今天能接 \u003Ca href=\"\u002Ftag\u002Fclaude\">Claude\u003C\u002Fa> Code，不代表明天所有工具都一樣順。誰能把設定維持一致，誰就更容易變成預設方案。\u003C\u002Fp>\u003Cp>我的判斷很直接。這類整合會越來越常見，但真正留下來的，不會是最多花招的那個，而是最少摩擦的那個。你如果正在選工具，先看接法，再看模型名，通常更準。\u003C\u002Fp>\u003Cp>\u003Ca href=\"\u002Fnews\u002Fllms-work-by-predicting-next-token-zh\">關鍵\u003C\u002Fa>字可以很簡單：先把工作流跑通，再談模型升級。這樣比較不會被一堆新名詞帶著走。\u003C\u002Fp>","三個多模態模型可透過 OpenAI 風格設定接入 Claude Code，也能搭配其他客戶端使用。重點在於統一介面、圖片與影片輸入，以及更容易切換模型。","zhuanlan.zhihu.com","https:\u002F\u002Fzhuanlan.zhihu.com\u002Fp\u002F2049943402922480335",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781892161193-9rla.png","tools","zh","b101c255-aaf5-4fb1-a6b3-b82bef35778f",[17,18,19,20,21,22,23],"Claude Code","多模態模型","OpenAI-style 設定","API","LLM","客戶端整合","Claude",[25,26,27,28],"OpenAI-style 設定讓多個客戶端更容易共用同一套模型接法。","三個多模態模型的價值，在於方便做橫向比較，不只是多一個選項。","Claude Code 的重點是工作流整合，不是單看模型名稱。","開發者應先看設定與切換成本，再看模型分數。",0,"2026-06-19T18:02:15.364369+00:00","2026-06-19T18:02:15.355+00:00","c3c88dd2-a940-438a-b359-0e5a24562273",{"tags":34,"relatedLang":42,"relatedPosts":46},[35,36,38,40],{"name":18,"slug":18},{"name":17,"slug":37},"claude-code",{"name":21,"slug":39},"llm",{"name":20,"slug":41},"api",{"id":15,"slug":43,"title":44,"language":45},"three-multimodal-models-work-in-claude-code-en","Three multimodal models now work in Claude Code","en",[47,53,59,65,71,77],{"id":48,"slug":49,"title":50,"cover_image":51,"image_url":51,"created_at":52,"category":13},"86316fab-2e99-4958-b79f-8c54ce44d5c4","ollama-turns-local-llms-into-copyable-setup-zh","Ollama 讓本地 LLM 變可抄配置","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781885006324-jvvo.png","2026-06-19T16:02:56.601682+00:00",{"id":54,"slug":55,"title":56,"cover_image":57,"image_url":57,"created_at":58,"category":13},"434cf6ed-e754-4dd7-be27-1aa0bc38761e","pypi-wasm-wheels-pyodide-zh","PyPI 開始收 WASM wheel，Pyodide 包裝順多了","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781875102614-0iu8.png","2026-06-19T13:17:56.091729+00:00",{"id":60,"slug":61,"title":62,"cover_image":63,"image_url":63,"created_at":64,"category":13},"45ae2326-cf6b-4164-883b-f33e48428502","windsurf-model-router-ide-opinion-june-2026-zh","Windsurf 正在變成模型路由器，而不只是 IDE","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781871473044-8win.png","2026-06-19T12:17:22.381+00:00",{"id":66,"slug":67,"title":68,"cover_image":69,"image_url":69,"created_at":70,"category":13},"d2a143b9-efa1-4ffd-adcb-7a315ae6344e","renesas-acquires-altium-pcb-design-tool-update-zh","瑞萨全资收购 Altium，PCB 教程更新","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781859766720-ow6s.png","2026-06-19T09:02:23.113145+00:00",{"id":72,"slug":73,"title":74,"cover_image":75,"image_url":75,"created_at":76,"category":13},"1e47b8fc-1eab-4342-83bd-a270d59a41f9","rust-forum-week-25-turns-ideas-into-shipping-work-zh","Rust 論壇週報把想法變交付","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781857111111-md5g.png","2026-06-19T08:18:04.893117+00:00",{"id":78,"slug":79,"title":80,"cover_image":81,"image_url":81,"created_at":82,"category":13},"300d082a-4df5-4a26-8b5b-7dff73dd0da3","claude-code-rust-native-terminal-interface-zh","Claude Code Rust 把終端機變輕了","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781854439295-lkeg.png","2026-06-19T07:33:29.722095+00:00",[84,89,94,99,104,109,114,119,124,129],{"id":85,"slug":86,"title":87,"created_at":88},"855cd52f-6fab-46cc-a7c1-42195e8a0de4","surepath-real-time-mcp-policy-controls-zh","SurePath 推出即時 MCP 政策控管","2026-03-26T07:57:40.77233+00:00",{"id":90,"slug":91,"title":92,"created_at":93},"9b19ab54-edef-4dbd-9ce4-a51e4bae4ebb","mcp-in-2026-the-ai-tool-layer-teams-use-zh","2026 年 MCP：團隊真的在用的 AI 工具層","2026-03-26T08:01:46.589694+00:00",{"id":95,"slug":96,"title":97,"created_at":98},"af9c46c3-7a28-410b-9f04-32b3de30a68c","prompting-in-2026-what-actually-works-zh","2026 提示工程，真正有用的是什麼","2026-03-26T08:08:12.453028+00:00",{"id":100,"slug":101,"title":102,"created_at":103},"05553086-6ed0-4758-81fd-6cab24b575e0","garry-tan-open-sources-claude-code-toolkit-zh","Garry Tan 開源 Claude Code 工具包","2026-03-26T08:26:20.068737+00:00",{"id":105,"slug":106,"title":107,"created_at":108},"042a73a2-18a2-433d-9e8f-9802b9559aac","github-ai-projects-to-watch-in-2026-zh","2026 必看 20 個 GitHub AI 專案","2026-03-26T08:28:09.619964+00:00",{"id":110,"slug":111,"title":112,"created_at":113},"a5f94120-ac0d-4483-9a8b-63590071ac6a","claude-code-vs-cursor-2026-zh","Claude Code 與 Cursor 深度對比：202…","2026-03-26T13:27:14.279193+00:00",{"id":115,"slug":116,"title":117,"created_at":118},"0975afa1-e0c7-4130-a20d-d890eaed995e","practical-github-guide-learning-ml-2026-zh","2026 機器學習入門 GitHub 實用指南","2026-03-27T01:16:49.712576+00:00",{"id":120,"slug":121,"title":122,"created_at":123},"bfdb467a-290f-4a80-b3a9-6f081afb6dff","aiml-2026-student-ai-ml-lab-repo-review-zh","AIML-2026：像課綱的學生實驗 Repo","2026-03-27T01:21:51.467798+00:00",{"id":125,"slug":126,"title":127,"created_at":128},"80cabc3e-09fc-4ff5-8f07-b8d68f5ae545","ai-trending-github-repos-and-research-feeds-zh","AI Trending：把 AI 資源收成一張表","2026-03-27T01:31:35.262183+00:00",{"id":130,"slug":131,"title":132,"created_at":133},"3ce6e6e2-bac5-463e-9f8d-45caabcc61f7","awesome-ai-for-science-research-tools-map-zh","AI 科研工具清單，開始像地圖了","2026-03-27T01:46:50.521945+00:00"]