OpenAI Decisions API Hits Public Beta With 10x Faster Typed Answers
https://www.marktechpost.com/2026/10/09/openai-decisions-api-hits-public-beta-with-10x-faster-typed-answers/📌 OpenAI推出Decisions API:讓模型直接吐出「程式碼能分支」的答案
TL;DR:新端點把LLM判斷轉成型別化答案,官方宣稱比Responses API快約10倍。
寫過 agent 或分類管線的人都懂這個痛點:你要的不是一段漂亮的文字,而是一個程式碼可以 if-else 的標籤,卻還是得先讓模型生成一整段 prose,再自己寫 parser 去抽取答案。OpenAI 這次直接把這個 pattern 做成了端點。
🤔 一個只回答「是非題」的端點
OpenAI Decisions API 已進入公開 Beta。它評估文字、圖片,或兩者皆有,回傳型別化的答案,而不是生成散文。目前唯一支援的模型是 gpt-6-luna,OpenAI 預期未來幾週內會正式 GA。
🧩 一個請求裡裝著共用證據與一串問題
一個 Decisions 請求只有 3 個欄位:model、input、questions。每個 question 都有自己的 name、type 與 instructions,而請求整體攜帶的是共用的 evidence(也就是 input)。回應則是一個 answers 陣列,用這些 name 做 key 對應。
OpenAI 用一個嚴重性分級的範例把機制講得很具體:假如某個嚴重度分類的三個等級機率分別是 0.1、0.7、0.2,加權後得到的分數是 1.1,正好落在「有變通方案(Workaround available)」與「完全阻斷(Fully blocked)」兩個等級之間,程式碼可以直接依這個連續分數做路由判斷,而不需要自己去解析模型吐出的文字再對應等級。
OpenAI 也劃清了三個端點的使用邊界:要機率、選項或分數,用 Decisions;要填自己的 JSON schema 或需要附帶解釋文字,用 Structured Outputs;要模型請求帶參數的工具呼叫,用 function calling。
📊 號稱快10倍,但沒有附上準確度數據
OpenAI 文件宣稱 Decisions API 比 Responses API 快約 10 倍。DevDay 現場揭露的數字則是:一次 decision 大約 150 毫秒,相較於一般 Luna 呼叫約 1.6 秒。計費上,gpt-6-luna 的輸入成本是每 100 萬 token 0.10 美元,沒有輸出、cache 讀取或 cache 寫入費用,但區域處理的溢價與長上下文倍率仍然適用——Luna 的 model card 顯示超過 272K token 的 prompt 會以 2 倍輸入費率計價。
值得留意的是,OpenAI 並未公布這個端點的準確度或校準(calibration)數據,文件建議開發者自己用標註過的範例來設定判斷門檻。
💡 市場上已經有更便宜的對手
最接近的競品是 2026 年 9 月 15 日上線的 TypeSafe Jev,一款 System One 模型,同樣回傳帶校準機率的型別化數值。Jev 的輸入定價是每 100 萬 token 0.042 美元,輸出免費,換算下來 OpenAI 的基礎費率大約是它的 2.4 倍。Jev 號稱端到端延遲在 70 到 500 毫秒(從美國西岸量測),支援最多 255 個選項,但目前仍在早期存取階段。OpenAI 的優勢則在於支援圖片輸入、提供合規選項,以及已經開放公開 Beta。
⚠️ 沒有準確度數據就是最大的風險
速度和價格都好比較,但一個決策端點最該被驗證的其實是它判斷得準不準、機率校準得好不好。OpenAI 目前沒有公開這類資料,意味著團隊在接上生產環境前,自己動手用標註資料驗證門檻設定幾乎是必要步驟。
🎯 實務啟示
如果你的 pipeline 裡還在用「prompt 模型生成文字 → regex 或字串比對抽取標籤」這種做法,Decisions API 提供的是一個結構上更直接的替代方案,尤其適合分類、評分、路由這類需要連續機率而非單一標籤的場景。但在導入前,務必自己驗證校準品質,而不是直接信任官方宣稱的速度數字。
🔗 來源
- 標題:OpenAI Decisions API Hits Public Beta With 10x Faster Typed Answers
- 作者/機構:Sana Hassan/MarkTechPost
- 連結:https://www.marktechpost.com/2026/10/09/openai-decisions-api-hits-public-beta-with-10x-faster-typed-answers/
#OpenAI #DecisionsAPI #LLM #StructuredOutputs #AIEngineering #API #MachineLearning #GPT #AIInfrastructure #FunctionCalling
原始資料 MarkTechPost · 收集於 2026-10-09
摘要原文
OpenAI has released the Decisions API in public beta. It turns text and images into typed answers your code can branch on. OpenAI team states the OpenAI Decisions API runs about 10x faster than the Responses API. It targets a common pattern: prompt an LLM, then parse its text into a label. The Decisions API is an OpenAI endpoint that evaluates text, images or both and returns typed answers. It does not generate prose. A request has 3 fields: model , input and questions . The only supported model today is gpt-6-luna . OpenAI expects general availability in the coming weeks . Each request carries shared evidence plus a list of questions. Each question has a unique name , a type and instructions. The response returns an answers array keyed by those names. OpenAI’s severity example makes the math concrete. Level probabilities of 0.1, 0.7 and 0.2 yield a score of 1.1. That value sits between ‘Workaround available’ and ‘Fully blocked’. OpenAI draws a clear line. Use Decisions for probabilities, choices or scores. Use Structured Outputs to fill your own JSON schema or write explanations. Use function calling when a model must request a tool call with arguments. OpenAI’s document claims about 10x faster responses than the Responses API. DevDay coverage put a decision near 150 ms, versus about 1.6 seconds for regular Luna calls. OpenAI has not published accuracy or calibration data for the endpoint. The docs advise setting thresholds with labeled examples from your own application. With gpt-6-luna , input costs $0.10 per 1M tokens . There are no output-token, cache-read or cache-write charges. Regional processing premiums and long-context multipliers still apply. Luna’s model card prices prompts above 272K tokens at 2x input rates. The closest rival is TypeSafe Jev , launched September 15, 2026. Jev is a System One model that returns typed values with calibrated probabilities. TypeSafe prices input at $0.042 per 1M tokens, with free output. That makes OpenAI’s base rate about 2.4x higher. TypeSafe reports 70 to 500 ms end-to-end, measured from the US West Coast. Jev supports up to 255 choices but remains in early access. OpenAI’s edge is image input, compliance options and open beta access. Check out the Technical details here . All credit goes to the researcher of this project. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter . Wait! are you on telegram? now you can join us on telegram as well. The post OpenAI Decisions API Hits Public Beta With 10x Faster Typed Answers appeared first on MarkTechPost .
由 claude-code/sonnet 自動生成