OrcaRouter Releases OrcaCyber Zero 1.5 Cybersecurity Model With 1M Context
https://www.marktechpost.com/2026/10/10/orcarouter-releases-orcacyber-zero-1-5-cybersecurity-model-with-1m-context/📌 OrcaCyber Zero 1.5：百萬 context 滲透測試模型上線
TL;DR：OrcaRouter 推出資安模型新版，主打漏洞複現與滲透測試，但效能數據全為廠商自報。
一款宣稱能自動複現 CVE、開發 exploit 的語言模型，聽起來像是資安界的噩夢，還是解方？端看是誰在用。
🤔 從 Zero 1.0 到 1.5，鎖定授權資安研究
OrcaRouter 於 2026 年 10 月 10 日發布 OrcaCyber Zero 1.5，接替今年 9 月 17 日上線的 OrcaCyber Zero 1.0。官方表示這是一款針對資安研究與授權資安工程進行 post-training 的模型，列出的用途包括漏洞複現（vulnerability reproduction）、exploit 開發、滲透測試、資安稽核與 cyber reasoning。模型本身具備 1M token 的 context window，並支援 native function calling 與結構化輸出（structured outputs）。
📊 四項廠商自報測試：Cybench、CVE-Bench、SWE-bench Pro V2
模型頁面列出 4 項「廠商自報」成績，最後更新時間為 2026 年 10 月 10 日：
- Cybench：涵蓋 40 道專業 CTF 題目，Orca 的 100% 成績實際涵蓋其中 39 題。
- CVE-Bench：基於 40 個 critical-severity 的 web CVE 建構，Orca 的 95.8% 涵蓋可評估的 24 題子集。
- SWE-bench Pro V2：官方特別註明此分數「不能直接與標準 SWE-bench Pro 的結果比較」。
值得留意的是，所有競品數據同樣來自各自廠商的自行公告，沒有任何一項經過獨立複現驗證。
🧩 OpenAI 相容 API，但存取被鎖在「資安研究」層級
開發者可將 base_url 設為 https://api.orcarouter.ai/v1，呼叫 orca/orcacyber-zero-1.5 使用模型，介面與 OpenAI API 相容。不過存取權限被限制在 Security Research tier，官方要求申請者提供 engagement（委託證明）、passkey 與簽署同意條款，目標族群鎖定受信任的資安研究者、紅隊與授權測試人員。
計價方面，輸入每 1M tokens 為 3 美元，輸出每 1M tokens 為 7.5 美元，cache read 則是每 1M tokens 0.75 美元。延遲數據顯示過去 7 天 p50 time-to-first-token 為 500 毫秒，p95 為 2.36 秒，但官方也坦承這個樣本只涵蓋 1.3K tokens 的流量，樣本量偏小；相較之下，Zero 1.0 在更大流量視窗下記錄的 p50 為 3.43 秒，兩者並非同等條件下的乾淨比較。
💡 延續性 post-training，而非架構創新
從公開資訊看，OrcaCyber Zero 1.5 更像是既有 Orca 模型線的 post-training 升級，鎖定垂直場景（漏洞複現、exploit 開發），而非全新的模型架構。其賣點在於把 context window 拉到 1M token，搭配 function calling 與結構化輸出，讓自動化的漏洞驗證與報告產出流程更順暢。對資安團隊而言，官方給出的訴求很直白：減少無效報告，提高驗證過與修復過的漏洞比例。
⚠️ 數字都來自廠商自己，無獨立覆核
所有列出的 benchmark 成績，包括 Cybench、CVE-Bench 與競品對比數字，都只是官方自行公布，沒有第三方重現。延遲比較也因樣本量與流量視窗不同而難以一對一對照。在採信這些宣稱前，仍需要獨立測試或第三方稽核佐證。
🎯 給資安團隊的實務提醒
若考慮導入此類模型進行授權滲透測試，建議先以內部 CVE 或已知弱點做小規模驗證，而非直接信任官方公布的百分比；同時留意存取權限審核流程（engagement、passkey、條款同意）可能拉長導入時間。
🔗 來源
- 標題：OrcaRouter Releases OrcaCyber Zero 1.5 Cybersecurity Model With 1M Context
- 作者／機構：Asif Razzaq, MarkTechPost
- 連結：https://www.marktechpost.com/2026/10/10/orcarouter-releases-orcacyber-zero-1-5-cybersecurity-model-with-1m-context/
#AI #Cybersecurity #LLM #PenetrationTesting #CVE #VulnerabilityResearch #RedTeam #AISecurity #ExploitDevelopment #CTF
原始資料 MarkTechPost · 收集於 2026-10-11
摘要原文
OrcaRouter has released OrcaCyber Zero 1.5 , a model for authorized vulnerability research. The model is the successor to OrcaCyber Zero 1.0 , which shipped on September 17, 2026. OrcaCyber Zero 1.5 is a post-trained Orca model for vulnerability reproduction, exploit development and penetration testing. It ships with a 1M-token context window, native function calling and structured outputs. For security teams, this model brings a simple message: fewer reports, more validated and fixed vulnerabilities . OrcaCyber Zero 1.5 is a frontier cybersecurity model and the successor to Zero 1.0 . Orca team states that it has post-trained for security research and authorized security engineering. Listed uses include vulnerability reproduction, exploit development, penetration testing, security auditing and cyber reasoning. The model page lists 4 vendor-reported results, last evaluated October 10, 2026: Cybench contains 40 professional CTF tasks, so the 100% covers 39 of them. CVE-Bench is built on 40 critical-severity web CVEs. Orca’s 95.8% covers a 24-task evaluable subset. The SWE-bench Pro V2 score is not directly comparable with standard SWE-bench Pro results. All competitor figures come from each vendor’s own announcement. None are independent replications. The model uses an OpenAI-compatible API . Developers set base_url to https://api.orcarouter.ai/v1 and call orca/orcacyber-zero-1.5 . Access is gated to the Security Research tier. Orca lists an engagement, a passkey and accepted terms as requirements. The tier targets trusted security researchers, red teams and authorized testing. Pricing is $3.00 per 1M input tokens and $7.50 per 1M output tokens. Cache reads cost $0.75 per 1M tokens. Over the past 7 days, p50 time-to-first-token was 500 ms and p95 was 2.36 s. That sample is small, at 1.3K tokens of traffic. Zero 1.0 recorded a 3.43 s p50 over a much larger traffic window. The 2 latency figures are not a clean comparison. Check out the model page , the launch post on X and the OrcaRouter docs . 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 OrcaRouter Releases OrcaCyber Zero 1.5 Cybersecurity Model With 1M Context appeared first on MarkTechPost .
由 claude-code/sonnet 自動生成