[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-5-open-source-llms-coding-cost-zh":3,"article-related-5-open-source-llms-coding-cost-zh":36,"series-industry-e642df22-7247-4581-91d5-0cec845a7269":88},{"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":28,"views":32,"created_at":33,"published_at":34,"topic_cluster_id":35},"e642df22-7247-4581-91d5-0cec845a7269","5-open-source-llms-coding-cost-zh","5 個開源 LLM：寫程式與成本","\u003Cp data-speakable=\"summary\">這份清單比較 5 個開源 \u003Ca href=\"\u002Ftag\u002Fllm\">LLM\u003C\u002Fa>，幫你快速選出最適合寫程式、\u003Ca href=\"\u002Ftag\u002F長上下文\">長上下文\u003C\u002Fa>或省成本的模型。\u003C\u002Fp>\u003Cp>\u003Ca href=\"\u002Fnews\u002Fvictor-wembanyama-news-vs-updates-vs-rumors-zh\">更新\u003C\u002Fa>到 \u003Ca href=\"\u002Fnews\u002Faeo-2026-financing-gaps-playbook-zh\">2026\u003C\u002Fa> 年 4 月後，你可以用這份排行在 5 個 open-weight 模型中做決策，先看整體品質，再看速度、上下文與價格，避免只靠印象挑模型。\u003C\u002Fp>\u003Ctable>\u003Cthead>\u003Ctr>\u003Cth>項目\u003C\u002Fth>\u003Cth>Quality Index\u003C\u002Fth>\u003Cth>Best Price\u003C\u002Fth>\u003Cth>Top Speed\u003C\u002Fth>\u003Cth>Max Context\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">Qwen3.7 Max\u003C\u002Fa>\u003C\u002Ftd>\u003Ctd>56.584\u003C\u002Ftd>\u003Ctd>$3.75\u002FM\u003C\u002Ftd>\u003Ctd>202 tok\u002Fs\u003C\u002Ftd>\u003Ctd>991K\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">Kimi K2.6\u003C\u002Fa>\u003C\u002Ftd>\u003Ctd>53.905\u003C\u002Ftd>\u003Ctd>$1.44\u002FM\u003C\u002Ftd>\u003Ctd>327 tok\u002Fs\u003C\u002Ftd>\u003Ctd>262K\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">MiMo-V2.5-Pro\u003C\u002Fa>\u003C\u002Ftd>\u003Ctd>53.829\u003C\u002Ftd>\u003Ctd>$1.20\u002FM\u003C\u002Ftd>\u003Ctd>88 tok\u002Fs\u003C\u002Ftd>\u003Ctd>1M\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">DeepSeek V4 Pro\u003C\u002Fa>\u003C\u002Ftd>\u003Ctd>51.509\u003C\u002Ftd>\u003Ctd>$0.54\u002FM\u003C\u002Ftd>\u003Ctd>159 tok\u002Fs\u003C\u002Ftd>\u003Ctd>1M\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">MiniMax-M2.7\u003C\u002Fa>\u003C\u002Ftd>\u003Ctd>49.615\u003C\u002Ftd>\u003Ctd>$0.52\u002FM\u003C\u002Ftd>\u003Ctd>446 tok\u002Fs\u003C\u002Ftd>\u003Ctd>205K\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003Ch2>1. Qwen3.7 Max：整體品質最佳\u003C\u002Fh2>\u003Cp>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">Qwen3.7 Max\u003C\u002Fa> 以 56.584 的 Quality Index 領先這份榜單，適合想先找「一個模型就能先上線」的團隊。它的 202 tok\u002Fs 與 991K context 也很實用，長提示、長文件與多輪工作都能撐得住。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779667574131-umho.png\" alt=\"5 個開源 LLM：寫程式與成本\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cul>\u003Cli>Best price: $3.75\u002FM\u003C\u002Fli>\u003Cli>Top speed: 202 tok\u002Fs\u003C\u002Fli>\u003Cli>Max context: 991K\u003C\u002Fli>\u003Cli>適合把品質放在第一順位的寫程式與推理工作\u003C\u002Fli>\u003C\u002Ful>\u003Cp>代價是價格偏高。如果你的流量大、預算緊，這顆不一定最省，但它最像「先求穩，再談優化」的選擇，特別適合評估期或核心產品流程。\u003C\u002Fp>\u003Ch2>2. Kimi K2.6：速度與供應來源最彈性\u003C\u002Fh2>\u003Cp>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">Kimi K2.6\u003C\u002Fa> 的 Quality Index 是 53.905，雖然不是第一名，但它以 327 tok\u002Fs 成為這組裡最快的模型，而且出現在 14 家供應商，部署與比價都更方便。\u003C\u002Fp>\u003Cul>\u003Cli>Best price: $1.44\u002FM\u003C\u002Fli>\u003Cli>Top speed: 327 tok\u002Fs\u003C\u002Fli>\u003Cli>Max context: 262K\u003C\u002Fli>\u003Cli>Providers: 14\u003C\u002Fli>\u003C\u002Ful>\u003Cp>如果你在意延遲、供應穩定性，或想在不同\u003Ca href=\"\u002Fnews\u002Fwhy-fragmented-data-breaks-cross-platform-performance-zh\">平台\u003C\u002Fa>間切換，這是很實際的選項。它的上下文不算最大，但對多數互動式應用已經夠用。\u003C\u002Fp>\u003Ch2>3. MiMo-V2.5-Pro：長文件與大上下文首選\u003C\u002Fh2>\u003Cp>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">MiMo-V2.5-Pro\u003C\u002Fa> 提供 1M \u003Ca href=\"\u002Ftag\u002Ftoken\">token\u003C\u002Fa> 的上下文窗口，最適合大型規格書、長篇研究、log 分析與多檔案程式任務。它的 Quality Index 53.829 也維持在高位，不會為了長上下文掉太多分數。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779667568289-7kjj.png\" alt=\"5 個開源 LLM：寫程式與成本\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cul>\u003Cli>Best price: $1.20\u002FM\u003C\u002Fli>\u003Cli>Top speed: 88 tok\u002Fs\u003C\u002Fli>\u003Cli>Max context: 1M\u003C\u002Fli>\u003Cli>適合文件密集、檔案很多的工作流\u003C\u002Fli>\u003C\u002Ful>\u003Cp>缺點是速度較慢，88 tok\u002Fs 不適合追求即時回應的高頻對話。但如果你的痛點是上下文不夠，這顆通常比再加複雜分段流程更簡單。\u003C\u002Fp>\u003Ch2>4. DeepSeek V4 Pro：大規模使用的低成本解\u003C\u002Fh2>\u003Cp>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">DeepSeek V4 Pro\u003C\u002Fa> 的價格只有 $0.54\u002FM，在這份清單裡屬於最省錢的主力之一。它的 Quality Index 51.509 不差，還有 159 tok\u002Fs 與 1M context，能兼顧成本與實用性。\u003C\u002Fp>\u003Cul>\u003Cli>Best price: $0.54\u002FM\u003C\u002Fli>\u003Cli>Top speed: 159 tok\u002Fs\u003C\u002Fli>\u003Cli>Max context: 1M\u003C\u002Fli>\u003Cli>適合重視單位成本的 production workload\u003C\u002Fli>\u003C\u002Ful>\u003Cp>如果你的用量大、預算固定，這種模型通常最容易算帳。它不是榜首，但在長上下文與低費率之間取得了很好的平衡，適合長期跑量。\u003C\u002Fp>\u003Ch2>5. MiniMax-M2.7：最便宜的高速選項\u003C\u002Fh2>\u003Cp>\u003Ca href=\"https:\u002F\u002Fwhatllm.org\u002F\">MiniMax-M2.7\u003C\u002Fa> 同時拿到最低價格 $0.52\u002FM 與最高速度 446 tok\u002Fs，對高頻聊天、路由、輕量助理很有吸引力。它的 Quality Index 49.615 較低，但仍可應付多數日常任務。\u003C\u002Fp>\u003Cul>\u003Cli>Best price: $0.52\u002FM\u003C\u002Fli>\u003Cli>Top speed: 446 tok\u002Fs\u003C\u002Fli>\u003Cli>Max context: 205K\u003C\u002Fli>\u003Cli>Providers: 6\u003C\u002Fli>\u003C\u002Ful>\u003Cp>這顆適合把「快」和「便宜」放在前面的人。若你要的是大量請求下的穩定吞吐，而不是最高 \u003Ca href=\"\u002Ftag\u002Fbenchmark\">benchmark\u003C\u002Fa>，它會是很務實的候選。\u003C\u002Fp>\u003Ch2>怎麼挑\u003C\u002Fh2>\u003Cp>想先選一顆全能型模型，就從 Qwen3.7 Max 開始；想要更快、供應商更多，選 Kimi K2.6；如果你的工作常常是長文件或長程式碼脈絡，\u003Ca href=\"\u002Ftag\u002Fmimo\">MiMo\u003C\u002Fa>-V2.5-Pro 最合適。要壓低成本，DeepSeek V4 Pro 和 MiniMax-M2.7 會更容易過預算。\u003C\u002Fp>\u003Cp>如果你是做本地部署或 Ollama 類型環境，硬體限制會直接影響體驗。這時候先看 VRAM、可接受的延遲與上下文需求，再用這份排行縮小候選名單。\u003C\u002Fp>","5 個開源 LLM 依寫程式、推理、速度、上下文與價格排序，附 2026 即時基準與選型建議。","whatllm.org","https:\u002F\u002Fwhatllm.org\u002Fbest-open-source-llm",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779667574131-umho.png","industry","zh","82a5471b-135f-4828-9534-6a11428045a2",[17,18,19,20,21,22,23,24,25,26,27],"open source LLM","coding model","context window","benchmark","token speed","cost","Qwen3.7 Max","Kimi K2.6","MiMo-V2.5-Pro","DeepSeek V4 Pro","MiniMax-M2.7",[29,30,31],"Qwen3.7 Max 最適合先求整體品質與穩定性。","Kimi K2.6 在速度與供應彈性上最突出。","MiMo-V2.5-Pro、DeepSeek V4 Pro 適合長上下文與控成本。",8,"2026-05-25T00:05:39.806482+00:00","2026-05-25T00:05:39.138+00:00","caa87b65-9bbc-46fe-bba8-4f4158dd2d8b",{"tags":37,"relatedLang":47,"relatedPosts":51},[38,39,41,43,45],{"name":20,"slug":20},{"name":17,"slug":40},"open-source-llm",{"name":21,"slug":42},"token-speed",{"name":19,"slug":44},"context-window",{"name":18,"slug":46},"coding-model",{"id":15,"slug":48,"title":49,"language":50},"5-open-source-llms-coding-cost-en","5 open source LLMs for coding and cost","en",[52,58,64,70,76,82],{"id":53,"slug":54,"title":55,"cover_image":56,"image_url":56,"created_at":57,"category":13},"173b8876-1867-4e0b-948f-27891d6b6364","openai-should-not-rush-its-ipo-just-to-win-the-ai-race-zh","OpenAI 不該為了搶 AI 賽道而急著 IPO","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781053365610-1hko.png","2026-06-10T01:02:19.886627+00:00",{"id":59,"slug":60,"title":61,"cover_image":62,"image_url":62,"created_at":63,"category":13},"3d7ff80a-4045-4b66-9e21-b6a8eb3b6f6d","openai-europe-privacy-policy-zh","OpenAI 歐洲隱私政策更新重點","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781052479369-yomr.png","2026-06-10T00:47:31.176745+00:00",{"id":65,"slug":66,"title":67,"cover_image":68,"image_url":68,"created_at":69,"category":13},"69002c63-177a-4723-9e63-d28506f08edd","openai-ads-sensitive-chats-policy-zh","OpenAI把廣告擋在敏感對話外是對的","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781051578409-en02.png","2026-06-10T00:32:23.404084+00:00",{"id":71,"slug":72,"title":73,"cover_image":74,"image_url":74,"created_at":75,"category":13},"ea98a8c9-ebe1-4258-8a2b-b0d82b25deed","ai-bootlegs-streaming-royalties-stick-figure-zh","AI bootlegs 正在抽走串流版稅","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781050681742-3rdh.png","2026-06-10T00:17:31.017287+00:00",{"id":77,"slug":78,"title":79,"cover_image":80,"image_url":80,"created_at":81,"category":13},"20d0b5fc-a363-481d-86b2-e30276a49e92","amd-microsoft-windows-ml-acceleration-zh","AMD 與 Microsoft 把 Windows ML 推進 GPU 與 N…","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781047980407-vd5p.png","2026-06-09T23:32:31.304436+00:00",{"id":83,"slug":84,"title":85,"cover_image":86,"image_url":86,"created_at":87,"category":13},"9a0692ba-a9c5-42eb-823d-8a0e6e6ae3fc","openai-ipo-filing-turns-hype-into-scrutiny-zh","OpenAI IPO 讓神話變審核","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781042614962-bj12.png","2026-06-09T22:03:04.524304+00:00",[89,94,99,104,109,114,119,124,129,134],{"id":90,"slug":91,"title":92,"created_at":93},"ee073da7-28b3-4752-a319-5a501459fb87","ai-in-2026-what-actually-matters-now-zh","2026 AI 真正重要的事","2026-03-26T07:09:12.008134+00:00",{"id":95,"slug":96,"title":97,"created_at":98},"83bd1795-8548-44c9-9a7e-de50a0923f71","trump-ai-framework-power-speech-state-preemption-zh","川普 AI 框架瞄準電力、言論與州權","2026-03-26T07:12:18.695466+00:00",{"id":100,"slug":101,"title":102,"created_at":103},"ea6be18b-c903-4e54-97b7-5f7447a612e0","nvidia-gtc-2026-big-ai-announcements-zh","NVIDIA GTC 2026 重點拆解","2026-03-26T07:14:26.62638+00:00",{"id":105,"slug":106,"title":107,"created_at":108},"4bcec76f-4c36-4daa-909f-54cd702f7c93","claude-users-spreading-out-and-getting-better-zh","Claude 用戶更分散，也更會用","2026-03-26T07:22:52.325888+00:00",{"id":110,"slug":111,"title":112,"created_at":113},"bd903b15-2473-4178-9789-b7557816e535","openclaw-raises-hard-question-for-ai-models-zh","OpenClaw 逼問 AI 模型價值","2026-03-26T07:24:54.707486+00:00",{"id":115,"slug":116,"title":117,"created_at":118},"eeac6b9e-ad9d-4831-8eec-8bba3f9bca6a","gap-google-gemini-checkout-fashion-search-zh","Gap 把結帳搬進 Gemini","2026-03-26T07:28:23.937768+00:00",{"id":120,"slug":121,"title":122,"created_at":123},"0740e53f-605d-4d57-8601-c10beb126f3c","google-pushes-gemini-transition-to-march-2026-zh","Google 把 Gemini 轉換延到 2026 年 3…","2026-03-26T07:30:12.825269+00:00",{"id":125,"slug":126,"title":127,"created_at":128},"e660d801-2421-4529-8fa9-86b82b066990","metas-llama-4-benchmark-scandal-gets-worse-zh","Meta Llama 4 分數風波又擴大","2026-03-26T07:34:21.156421+00:00",{"id":130,"slug":131,"title":132,"created_at":133},"183f9e7c-e143-40bb-a6d5-67ba84a3a8bc","accenture-mistral-ai-sovereign-enterprise-deal-zh","Accenture 攜手 Mistral AI 賣主權 AI","2026-03-26T07:38:14.818906+00:00",{"id":135,"slug":136,"title":137,"created_at":138},"191d9b1b-768a-478c-978c-dd7431a38149","mistral-ai-faces-its-hardest-year-yet-zh","Mistral AI 迎來最硬的一年","2026-03-26T07:40:23.716374+00:00"]