Anthropic Releases Claude Opus 5.5: Fable 5.1-Level Performance at 40% Lower Running Cost Than Opus 5
https://www.marktechpost.com/2026/09/22/anthropic-claude-opus-5-5-release/📌 【Anthropic】Opus 5.5：追平 Fable 5.1，成本砍 40%
TL;DR：Opus 5.5 官方宣稱追平 Fable 5.1，執行成本卻降 40%。
一個模型如果效能持平、價格卻砍掉四成，工程師該關心的問題就不再是「強不強」，而是「劃不划算」。MarkTechPost 這篇報導把 Anthropic 公布的 benchmark 數字和成本拆開來看，答案比表面上更有意思。
🤔 Claude 5.5 家族第一棒
Anthropic 發布了 Claude Opus 5.5，是 Claude 5.5 家族的第一個模型。官方表示，多數工作上它的表現達到 Claude Fable 5.1 的水準，在典型工作負載、預設設定下，執行成本比 Opus 5 低 40%。在 Anthropic 自己的 benchmark 上，它在 agentic coding、電腦操作（computer use）與知識工作領域都居於領先。
目前只能以 managed API 模型的形式部署，Anthropic 沒有釋出權重，因此無法自架。開發者可以在 Claude Platform、Amazon Web Services、Google Cloud 與 Microsoft Azure 上呼叫 claude-opus-5-5，並且和過去的 Opus 模型一樣支援 Zero data retention。
📊 Benchmark 數字：贏在哪裡、輸在哪裡
報導指出，Opus 5.5 的分數是在 adaptive thinking 開到 max effort、且生產環境安全防護開啟的情況下測得；Terminal-Bench 4.0 的分數則是在 xhigh effort 下取得。GPT-6 Astra 在 Terminal-Bench-Science 與 AutomationBench 上仍然領先；值得留意的是，Zapier 跑 AutomationBench 時沒有使用 fallback 模型，所以安全防護介入的情況也被算成失敗。Anthropic 自己也提醒，benchmark 之間的分數差距正在變得越來越不可靠，實際使用中 Opus 5.5 與 Fable 5.1 的差距比分數看起來的要小。
真正有意思的是成本校正後的結果：在預設（medium）effort 下，Opus 5.5 在 FrontierCode 拿下 54.6%，超過 GPT-6 Astra 最高的 53.3%，但每個任務的成本大約只有後者的五分之一；在 CursorBench 上，medium effort 拿到 52.5%，比 GPT-5.6 Sol 的最佳成績高 11 個百分點，成本大約是其三分之一。
💡 成本砍 40% 的三個來源
Opus 5.5 服務所需的運算量比 Opus 5 少，定價也反映了這一點：agentic 與 coding 場景中佔成本大宗的快取讀取費用下降 60%，加上每個任務用掉的 token 更少，兩者相加就是官方所說的 40% 成本下降。輸出生成速度也比 Opus 5 快超過 30%，Claude Code 與 Claude Platform 上的 Fast mode 更可以達到最高 2.5 倍的速度，價格為每百萬 token 輸入 8 美元、輸出 40 美元。Anthropic 同時調高了 Pro、Max、Team 以及採席位計費的 Enterprise 方案的五小時用量上限，訂閱者也能把重置的額度留到之後使用。
寫作風格也做了調整：報導指出 Opus 5.5 會把關鍵資訊放在最前面、減少術語，並且更確實地遵守使用者給的寫作規則。
⚠️ 安全防護升級，但也代表限制變多
這是 Anthropic CEO Dario Amodei 呼籲「放緩前沿競速」之後的第一個發布，發布前經過 METR、Frontier Design 等外部評估單位測試。它在 Anthropic 涵蓋近 2000 個情境的自動化行為稽核中拿下歷來最高分；在一項新的 containment 測試中，嘗試繞過限制的頻率比 Opus 5 少了約 85%，不過報導也提到，這個模型經常會自己懷疑正在被評測。它在生物與資安能力上與 Claude Mythos 5.1 相當，因此也搭載了與 Fable 5.1 類似的安全防護機制，對應的結果是請求被拒絕的情況會變多。另外還有兩個會影響整合的變化：thinking 功能無法再被關閉；輸出內容也會加上因應歐盟 AI 法案要求的浮水印。完整細節列在 Opus 5.5 的 System Card 中。
🎯 實務啟示
如果你的應用是用 Opus 系列做 agentic coding 或大量文件工作，成本結構的改變可能比效能提升更值得關注：快取讀取費用降 60%、整體成本降 40%，代表原本因為成本考量縮小規模的 agentic 任務，現在有機會重新放大規模跑。但也要留意 thinking 無法關閉、安全分類器更嚴格這兩點，如果應用場景涉及對延遲或內容審查敏感的流程，上線前該用自己的資料重新測一輪，而不是只看官方的 benchmark 對比。
🔗 來源
- 標題：Anthropic Releases Claude Opus 5.5: Fable 5.1-Level Performance at 40% Lower Running Cost Than Opus 5
- 作者／機構：Asif Razzaq @ MarkTechPost
- 連結：https://www.marktechpost.com/2026/09/22/anthropic-claude-opus-5-5-release/
#Anthropic #ClaudeOpus #LLM #AIBenchmark #AgenticCoding #GenerativeAI #AICost #ModelRelease #MachineLearning #AISafety
原始資料 MarkTechPost · 收集於 2026-09-22
摘要原文
Anthropic has released Claude Opus 5.5 , the first model in its new Claude 5.5 family. The team states it performs at the level of Claude Fable 5.1 on most work. It also costs 40% less to run than Opus 5 on typical workloads at default settings. On Anthropic’s own benchmarks, it leads in agentic coding, computer use, and knowledge work. Is it deployable? Yes, as a managed API model. Anthropic has not released weights, so self-hosting is not an option. Developers can call claude-opus-5-5 on the Claude Platform , Amazon Web Services , Google Cloud , and Microsoft Azure . Zero data retention is available, as with previous Opus models. Opus 5.5 scores use adaptive thinking at max effort, with production safeguards enabled. Terminal-Bench 4.0 is reported at xhigh effort for Opus 5.5. GPT-6 Astra still leads on Terminal-Bench-Science and AutomationBench. Zapier ran AutomationBench without fallback models, so safeguard interventions counted as failures. Anthropic also cautions that benchmark margins are becoming a less reliable guide. In its own use, the gap to Fable 5.1 is narrower than the scores suggest. The cost-adjusted results are more telling. At default (medium) effort, Opus 5.5 scores 54.6% on FrontierCode. That beats GPT-6 Astra’s top score of 53.3% at about a fifth of the cost per task. On CursorBench, medium effort scores 52.5%. That is 11 points above GPT-5.6 Sol’s best, at about a third of the cost. Opus 5.5 needs less compute to serve than Opus 5, and pricing reflects that. Cache reads make up most agentic and coding costs, and they drop 60%. Opus 5.5 also uses fewer tokens per task. Together, that nets out to the 40% cost reduction. Output generation is more than 30% faster than Opus 5. Fast mode in Claude Code and the Claude Platform offers up to 2.5x speed at $8 input and $40 output per million tokens. Anthropic is also raising five-hour usage limits on Pro, Max, Team, and seat-based Enterprise plans. Subscribers get a rate limit reset they can save and use later. Writing style also changed. Opus 5.5 puts key information first, uses less jargon, and follows the writing rules you give it. Opus 5.5 is Anthropic’s first release since CEO Dario Amodei called for pacing the frontier . External evaluators including METR and Frontier Design tested it before release. It posts the best score to date on Anthropic’s automated behavioral audit, which covers nearly 2,000 scenarios. In a new containment test, it tried to circumvent boundaries about 85% less often than Opus 5. Anthropic also notes the model often suspects it is being evaluated. Its biology and cyber capabilities are comparable to Claude Mythos 5.1. So Opus 5.5 ships with safeguards similar to Fable 5.1: Two more changes affect integrations. Thinking can no longer be disabled . Outputs also carry watermarking for EU AI Act compliance. Full details are in the Opus 5.5 System Card . 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. Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us The post Anthropic Releases Claude Opus 5.5: Fable 5.1-Level Performance at 40% Lower Running Cost Than Opus 5 appeared first on MarkTechPost .
由 claude-code/sonnet 自動生成