[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-claude-mythos-preview-benchmarks-price-context-zh":3,"article-related-claude-mythos-preview-benchmarks-price-context-zh":33,"series-model-release-fe00f162-ac54-4c4e-b2db-c8dd5f2b53a7":84},{"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":25,"views":29,"created_at":30,"published_at":31,"topic_cluster_id":32},"fe00f162-ac54-4c4e-b2db-c8dd5f2b53a7","claude-mythos-preview-benchmarks-price-context-zh","Claude Mythos Preview：價格、上下文與定位","\u003Cp data-speakable=\"summary\">\u003Ca href=\"\u002Ftag\u002Fclaude-mythos\">Claude Mythos\u003C\u002Fa> Preview 是 \u003Ca href=\"\u002Ftag\u002Fanthropic\">Anthropic\u003C\u002Fa> 內部測試中的多模態模型，定位高於 Opus，外界現在最想知道的是價格、上下文長度和實際能力。\u003C\u002Fp>\u003Cp>說真的，這名字一出來，社群就開始腦補了。\u003Ca href=\"https:\u002F\u002Fllm-stats.com\u002Fmodels\u002Fclaude-mythos-preview\" target=\"_blank\" rel=\"noopener\">LLM Stats\u003C\u002Fa> 先把它掛上去，大家就開始看規格、猜價格、翻 benchmark。\u003C\u002Fp>\u003Cp>目前能確認的資訊不多，但夠讓人畫\u003Ca href=\"\u002Fnews\u002F5-takeaways-from-the-indiana-fever-schedule-zh\">重點\u003C\u002Fa>。它標成 \u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002Fclaude\" target=\"_blank\" rel=\"noopener\">Anthropic\u003C\u002Fa> 出品、支援文字和圖片、而且是專有授權。再加上它被放在 \u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002Fclaude\u002Fopus\" target=\"_blank\" rel=\"noopener\">Claude Opus\u003C\u002Fa> 上面，這就不是一般小更新了。\u003C\u002Fp>\u003Ctable>\u003Cthead>\u003Ctr>\u003Cth>欄位\u003C\u002Fth>\u003Cth>內容\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd>模型名稱\u003C\u002Ftd>\u003Ctd>Claude Mythos Preview\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>公司\u003C\u002Ftd>\u003Ctd>Anthropic\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>授權\u003C\u002Ftd>\u003Ctd>專有\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>輸入型態\u003C\u002Ftd>\u003Ctd>文字、圖片\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>定位\u003C\u002Ftd>\u003Ctd>高於 Opus\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>內部代號\u003C\u002Ftd>\u003Ctd>Capybara\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003Ch2>這個列表到底透露什麼\u003C\u002Fh2>\u003Cp>先講白了，這份外部列表不是\u003Ca href=\"\u002Fnews\u002Findiana-fever-news-vs-wnba-team-page-zh\">官方\u003C\u002Fa>公告。它比較像是模型資料站先把線索拼起來，讓大家看到一個輪廓。輪廓雖然薄，但還是有用。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779401175670-k2yg.png\" alt=\"Claude Mythos Preview：價格、上下文與定位\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>第一個重點是多模態。這代表它不是只會吃文字。圖片輸入通常會牽涉 OCR、視覺理解、圖表判讀，這些都會直接影響產品可用性。\u003C\u002Fp>\u003Cp>第二個重點是專有授權。這表示你現在不用想開源權重，也不用想自己架一套就能拿來改。對開發者來說，這種模型通常就是走 API 路線。\u003C\u002Fp>\u003Cul>\u003Cli>來自 Anthropic\u003C\u002Fli>\u003Cli>支援文字與圖片\u003C\u002Fli>\u003Cli>採專有授權\u003C\u002Fli>\u003Cli>定位高於 Opus\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>為什麼 benchmark 會讓人緊張\u003C\u002Fh2>\u003Cp>外流資訊裡最吸睛的，是它被寫成能找出大量 zero-day 漏洞。這種說法很猛，但也很容易被誇大。因為「找到漏洞」和「能穩定找漏洞」是兩回事。\u003C\u002Fp>\u003Cp>如果模型真的在安全測試裡表現很強，那代表它可能在推理、程式碼分析、工具使用上都有進步。這對資安團隊很有吸引力，但也代表風險更高。模型越會找洞，越需要管控。\u003C\u002Fp>\u003Cp>Anthropic 一直很愛講安全與可控部署。這不是口號問題，而是商業問題。你要把模型賣給企業，就得讓法務、資安、採購都點頭。\u003C\u002Fp>\u003Cblockquote>“We’re not building models to be clever in a demo; we’re building them to be useful in the real world.” — Dario Amodei，Anthropic 執行長，\u003Ca href=\"https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fanthropic-dario-amodei-ai-safety\u002F\" target=\"_blank\" rel=\"noopener\">WIRED\u003C\u002Fa>\u003C\u002Fblockquote>\u003Cp>這句話放在這裡很貼切。因為 Mythos Preview 目前就是一個「看起來很強，但還沒正式上線」的狀態。能不能用，才是重點。\u003C\u002Fp>\u003Ch2>跟 Anthropic 現有模型怎麼比\u003C\u002Fh2>\u003Cp>現在大家熟的是 \u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002Fclaude\u002Fopus\" target=\"_blank\" rel=\"noopener\">Claude Opus\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002Fclaude\u002Fsonnet\" target=\"_blank\" rel=\"noopener\">Claude Sonnet\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002Fclaude\u002Fhaiku\" target=\"_blank\" rel=\"noopener\">Claude Haiku\u003C\u002Fa>。這三個名字已經很清楚地切出不同價位和用途。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779401174497-z016.png\" alt=\"Claude Mythos Preview：價格、上下文與定位\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>Mythos Preview 沒有公開價格頁，也沒有官方 API 文件。這點很重要。因為對開發者來說，模型不是只看分數，還要看每百萬 \u003Ca href=\"\u002Ftag\u002Ftoken\">token\u003C\u002Fa> 成本、延遲、rate limit、以及上下文長度。\u003C\u002Fp>\u003Cp>如果它真的在 Opus 之上，那 Anthropic 很可能想把它放在更高價位。這種策略很常見。模型越強，通常越貴。問題只在於，貴多少，值不值。\u003C\u002Fp>\u003Cul>\u003Cli>Opus、Sonnet、Haiku 都已公開\u003C\u002Fli>\u003Cli>Mythos Preview 還沒有公開價格\u003C\u002Fli>\u003Cli>官方沒有列出 context window\u003C\u002Fli>\u003Cli>也沒有公開 API 文件\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>開發者最該盯的三個數字\u003C\u002Fh2>\u003Cp>我覺得接下來最重要的，不是代號，而是三個數字。第一個是價格。第二個是 context window。第三個是延遲。這三個數字會直接決定它能不能進產品。\u003C\u002Fp>\u003Cp>如果價格比 Opus 高很多，但效果只多一點，那多數團隊不會買單。反過來，如果它能在長文件摘要、程式碼審查、客服代理上省掉大量人工，那就有機會。\u003C\u002Fp>\u003Cp>對台灣團隊來說，最實際的場景通常是三種：內部知識庫、客服自動化、以及軟體開發輔助。這三種都很吃上下文長度，也很吃穩定性。\u003C\u002Fp>\u003Cul>\u003Cli>價格會影響導入意願\u003C\u002Fli>\u003Cli>上下文長度決定能處理多少資料\u003C\u002Fli>\u003Cli>延遲影響產品體驗\u003C\u002Fli>\u003Cli>API 穩定性影響上線風險\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>多模態現在已經不是加分題\u003C\u002Fh2>\u003Cp>以前大家看到圖片輸入，會覺得是附加功能。現在不是了。很多企業資料本來就不是純文字。發票、截圖、報表、流程圖，通通是圖片。\u003C\u002Fp>\u003Cp>這也是為\u003Ca href=\"\u002Fnews\u002Fwei-shen-me-2026-indiana-fever-jiu-shi-yao-xian-zai-ying-zh-zh\">什麼\u003C\u002Fa>多模態模型越來越像基本配備。你如果只能吃文字，很多工作流程就得先轉檔、先 OCR、先清洗，整個管線都變長。\u003C\u002Fp>\u003Cp>Anthropic 如果把 Mythos Preview 做成更高階的多模態模型，那它瞄準的就不只是聊天機器人，而是更完整的工作代理。這一點比名字本身重要多了。\u003C\u002Fp>\u003Ch2>市場脈絡也很簡單\u003C\u002Fh2>\u003Cp>\u003Ca href=\"\u002Ftag\u002Fopenai\">OpenAI\u003C\u002Fa>、\u003Ca href=\"\u002Ftag\u002Fgoogle\">Google\u003C\u002Fa>、Anthropic 這幾家現在都在拼模型分層。公開層級會有比較清楚的價格和文件，內部高階模型則常常先拿去測試，再決定要不要放出來。\u003C\u002Fp>\u003Cp>這種做法很合理。因為高階模型一旦開放，成本、風險、客服壓力都會一起上來。尤其是資安與企業場景，模型一強，出事也會更大條。\u003C\u002Fp>\u003Cp>所以 Mythos Preview 目前比較像訊號，不像產品。它告訴我們 Anthropic 還在往更高階的模型線走，但真正能不能用，還得看後續的 API、價格和限制。\u003C\u002Fp>\u003Ch2>我會怎麼看這件事\u003C\u002Fh2>\u003Cp>講白了，這東西現在最像一張預告圖。你知道它存在，也知道它可能很強，但你還不能拿來上線。對工程團隊來說，這種時候先別衝動。\u003C\u002Fp>\u003Cp>我會建議先觀察三件事。第一，Anthropic 何時公開。第二，價格怎麼訂。第三，context window 有沒有拉長到能吃大文件和長對話。如果這三項都漂亮，才值得認真評估。\u003C\u002Fp>\u003Cp>如果你是開發者，現在最實際的做法不是猜，而是先把自己的工作流整理好。等官方文件出來，你就能立刻測試，而不是在社群留言裡瞎猜。\u003C\u002Fp>\u003Cp>如果你想追後續，我會盯 Anthropic 官方公告、API 文件和價格頁。這些東西一出來，才是真正能決定要不要導入的訊號。\u003C\u002Fp>","Anthropic 的 Claude Mythos Preview 目前只在外部名單露面，主打多模態、位階高於 Opus，外界最在意的是價格、上下文長度與是否真能落地。","llm-stats.com","https:\u002F\u002Fllm-stats.com\u002Fmodels\u002Fclaude-mythos-preview",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1779401175670-k2yg.png","model-release","zh","82f156ee-9163-4fec-a11b-19a123329980",[17,18,19,20,21,22,23,24],"Claude Mythos Preview","Anthropic","Claude Opus","多模態模型","上下文長度","API 價格","LLM","AI 模型",[26,27,28],"Claude Mythos Preview 目前只在外部名單露面，還不是可直接使用的正式產品。","它被標成多模態，而且定位高於 Opus，代表 Anthropic 可能在測更高階模型。","開發者最該等的不是代號，而是價格、context window、延遲和 API 穩定性。",9,"2026-05-21T22:05:49.565724+00:00","2026-05-21T22:05:49.386+00:00","0ccb5d2e-69f1-4354-a3e0-cb370221cd95",{"tags":34,"relatedLang":43,"relatedPosts":47},[35,37,38,40,42],{"name":17,"slug":36},"claude-mythos-preview",{"name":20,"slug":20},{"name":19,"slug":39},"claude-opus",{"name":18,"slug":41},"anthropic",{"name":21,"slug":21},{"id":15,"slug":44,"title":45,"language":46},"claude-mythos-preview-benchmarks-price-context-en","Claude Mythos Preview: Benchmarks, Price, Context","en",[48,54,60,66,72,78],{"id":49,"slug":50,"title":51,"cover_image":52,"image_url":52,"created_at":53,"category":13},"466021f3-b8a4-4ecb-ad64-8070beaf9cbc","gemini-1-5-pro-002-flash-002-2-0-flash-update-zh","Gemini 1.5 與 2.0 Flash 更新上線","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780999389960-97qh.png","2026-06-09T10:02:27.849751+00:00",{"id":55,"slug":56,"title":57,"cover_image":58,"image_url":58,"created_at":59,"category":13},"66ce4542-3c93-4a0c-ab52-5e6f90a36212","minimax-m3-kai-fang-quan-zhong-xie-cheng-shi-reng-neng-ying-zh","MiniMax M3 證明開放權重在寫程式上仍能贏","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780968786191-lele.png","2026-06-09T01:32:30.829528+00:00",{"id":61,"slug":62,"title":63,"cover_image":64,"image_url":64,"created_at":65,"category":13},"948a7dc4-b172-42f9-9bef-abcbbffaca18","gemini-35-flash-pricing-benchmarks-zh","Gemini 3.5 Flash 價格與長上下文解析","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780840978961-6b9n.png","2026-06-07T14:02:29.835438+00:00",{"id":67,"slug":68,"title":69,"cover_image":70,"image_url":70,"created_at":71,"category":13},"5507f140-5223-4f68-ade6-30d9e5457638","gemma-4-12b-specs-benchmarks-run-locally-zh","怎麼做 Gemma 4 12B 本地部署","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780777971165-4bit.png","2026-06-06T20:32:24.857611+00:00",{"id":73,"slug":74,"title":75,"cover_image":76,"image_url":76,"created_at":77,"category":13},"ef42a437-8b06-4ff5-a135-ece7662c01f4","best-kimi-models-2026-k2-5-vs-k2-thinking-zh","2026 最佳 Kimi 模型：K2.5 對 K2 Thinking","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780770790333-x3lk.png","2026-06-06T18:32:39.410186+00:00",{"id":79,"slug":80,"title":81,"cover_image":82,"image_url":82,"created_at":83,"category":13},"fd2ad557-5c09-4758-964d-cda1c3c87a4c","kimi-k2-6-open-source-coding-agent-swarm-zh","Kimi K2.6 開源加上 Agent Swarm","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780761795960-0zg9.png","2026-06-06T16:02:21.702099+00:00",[85,90,95,100,105,110,115,120,125,130],{"id":86,"slug":87,"title":88,"created_at":89},"58b64033-7eb6-49b9-9aab-01cf8ae1b2f2","nvidia-rubin-six-chips-one-ai-supercomputer-zh","NVIDIA Rubin 把六顆晶片塞進 AI 機櫃","2026-03-26T07:18:45.861277+00:00",{"id":91,"slug":92,"title":93,"created_at":94},"0dcc2c61-c2a6-480d-adb8-dd225fc68914","march-2026-ai-model-news-what-mattered-zh","2026 年 3 月 AI 模型新聞重點","2026-03-26T07:32:08.386348+00:00",{"id":96,"slug":97,"title":98,"created_at":99},"214ab08b-5ce5-4b5c-8b72-47619d8675dd","why-small-models-are-winning-on-device-ai-zh","小模型為何吃下裝置端 AI","2026-03-26T07:36:30.488966+00:00",{"id":101,"slug":102,"title":103,"created_at":104},"785624b2-0355-4b82-adc3-de5e45eecd88","midjourney-v8-faster-images-higher-costs-zh","Midjourney V8 變快了，也變貴了","2026-03-26T07:52:03.562971+00:00",{"id":106,"slug":107,"title":108,"created_at":109},"cda76b92-d209-4134-86c1-a60f5bc7b128","xiaomi-mimo-trio-agents-robots-voice-zh","小米 MiMo 三模型瞄準代理、機器人與語音","2026-03-28T03:05:08.779489+00:00",{"id":111,"slug":112,"title":113,"created_at":114},"9e1044b4-946d-47fe-9e2a-c2ee032e1164","xiaomi-mimo-v2-pro-1t-moe-agents-zh","小米 MiMo-V2-Pro 登場：1T MoE 模型","2026-03-28T03:06:19.002353+00:00",{"id":116,"slug":117,"title":118,"created_at":119},"c4b6186f-bd84-4598-997e-c6e31d543c0d","cursor-composer-2-agentic-coding-model-zh","Cursor Composer 2 走向代理式寫碼","2026-03-28T03:13:06.422716+00:00",{"id":121,"slug":122,"title":123,"created_at":124},"e112e76f-ec3b-408f-810e-e93ae21a888a","apple-siri-gemini-distilled-models-zh","Apple Siri 牽手 Gemini 的真相","2026-03-29T04:52:57.886544+00:00",{"id":126,"slug":127,"title":128,"created_at":129},"c679b51f-194a-463b-87fc-7695256ff752","mimo-v2-pro-vs-omni-vs-flash-2026-zh","MiMo V2 Pro、Omni、Flash 怎麼選","2026-04-02T01:18:43.576128+00:00",{"id":131,"slug":132,"title":133,"created_at":134},"3b988fd7-6749-4f01-ba25-c0ad7486dc31","z-ai-glm-5v-turbo-design2code-claude-zh","GLM-5V-Turbo 在 Design2Code 贏了…","2026-04-02T04:03:36.31741+00:00"]