{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-23T06:40:50.084Z","headline":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于漏洞定位","description":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于在代码库中定位已知漏洞。Antares-1B 在 VLoc Bench 上达到 0.209 File F1，超过 753B 参数的 GLM-5.2（0.186）。Antares-350M 和 Antares-1B 以 Apache 2.0 许可证开源。","url":"https://www.aioga.com/news/cmrvqaiiw00csbi85v6yivotz/","mainEntityOfPage":"https://www.aioga.com/news/cmrvqaiiw00csbi85v6yivotz/","datePublished":"2026-07-22T06:27:02.000Z","dateModified":"2026-07-22T06:27:02.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://www.marktechpost.com/2026/07/21/cisco-foundation-ai-releases-antares-350m-and-1b-open-weight-models-that-localize-known-vulnerabilities-inside-real-codebases","https://aihot.virxact.com/items/cmrvqaiiw00csbi85v6yivotz"],"canonicalUrl":"https://www.aioga.com/news/cmrvqaiiw00csbi85v6yivotz/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于在代码库中定位已知漏洞。 Aioga 将其归入「模型更新」方向，重点关注它对真实使用和行业竞争的影响。","url":"https://www.aioga.com/news/cmrvqaiiw00csbi85v6yivotz/","dateCreated":"2026-07-22T06:27:02.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"marktechpost.com source article","url":"https://www.marktechpost.com/2026/07/21/cisco-foundation-ai-releases-antares-350m-and-1b-open-weight-models-that-localize-known-vulnerabilities-inside-real-codebases","datePublished":"2026-07-22T06:27:02.000Z","provider":{"@type":"Organization","name":"marktechpost.com","url":"https://www.marktechpost.com/2026/07/21/cisco-foundation-ai-releases-antares-350m-and-1b-open-weight-models-that-localize-known-vulnerabilities-inside-real-codebases"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmrvqaiiw00csbi85v6yivotz","datePublished":"2026-07-22T06:27:02.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmrvqaiiw00csbi85v6yivotz"}}],"aggregationSource":"MarkTechPost（RSS）","originalPublisher":{"name":"marktechpost.com","url":"https://www.marktechpost.com/2026/07/21/cisco-foundation-ai-releases-antares-350m-and-1b-open-weight-models-that-localize-known-vulnerabilities-inside-real-codebases"},"article":{"id":"cmrvqaiiw00csbi85v6yivotz","slug":"cmrvqaiiw00csbi85v6yivotz","url":"https://www.aioga.com/news/cmrvqaiiw00csbi85v6yivotz/","title":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于漏洞定位","title_en":"Cisco Foundation AI Releases Antares： 350M and 1B Open-Weight Models That Localize Known Vulnerabilities Inside Real Codebases","summary":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于在代码库中定位已知漏洞。Antares-1B 在 VLoc Bench 上达到 0.209 File F1，超过 753B 参数的 GLM-5.2（0.186）。Antares-350M 和 Antares-1B 以 Apache 2.0 许可证开源。","source":"MarkTechPost（RSS）","sourceUrl":"https://www.marktechpost.com/2026/07/21/cisco-foundation-ai-releases-antares-350m-and-1b-open-weight-models-that-localize-known-vulnerabilities-inside-real-codebases","aiHotUrl":"https://aihot.virxact.com/items/cmrvqaiiw00csbi85v6yivotz","publishedAt":"2026-07-22T06:27:02.000Z","category":"模型更新","score":59,"selected":false,"articleBody":["Cisco Foundation AI：https://blogs.cisco.com/ai/introducing-antares-the-most-efficient-open-weight-ai-models-for-vulnerability-localization has released Antares , a family of security small language models (SLMs) built for one narrow security task. The task is vulnerability localization. Given a vulnerability description and a repository, find the files containing the flaw.","Two models are open-weight and available now on Hugging Face, Antares-350M：https://huggingface.co/fdtn-ai/antares-350m and Antares-1B：https://huggingface.co/fdtn-ai/antares-1b. Both are Apache 2.0. Cisco team also shipped the Vulnerability Localization Benchmark：https://cisco-foundation-ai.github.io/vulnerability-localization-benchmark/ (VLoc Bench), a 500-task agentic evaluation, under the same license.","The main result is not a new state of the art. It is that a 1B model reaches 0.209 File F1. GPT-5.5 reaches 0.229, and a 753B open-weight model reaches 0.186.","Software security depends on connecting external vulnerability knowledge to internal source code. That knowledge lives in public databases, advisories, and Common Weakness Enumerations：https://cwe.mitre.org/. The code lives in repositories that are large, modular, and dependency-rich.","Connecting the two is expensive. Devs search unfamiliar code, follow naming conventions, inspect call paths, and compare candidate files. Cisco’s framing is that this first triage step is where the cost concentrates.","Antares does not replace the application security toolchain. Cisco is explicit about this. Dev teams still need dependency scanning, secret scanning, dynamic testing, container checks, threat modeling, and expert review.","Antares consists of three decoder-only transformers at 350M, 1B, and 3B parameters. All three initialize from IBM Granite 4.0：https://huggingface.co/ibm-granite/granite-4.0-1b checkpoints. They share a tokenizer and architecture: grouped-query attention, SwiGLU MLPs, RMSNorm, RoPE, and shared input/output embeddings.","Antares is not evaluated as a standalone sequence model. It runs inside a constrained loop with three tools.","The model receives a CWE category description and nothing else. No advisory text, no file hints, no severity details. It then issues read-only terminal commands against a Docker sandbox with networking disabled. Command output is truncated to 2,000 characters before entering the transcript.","The budget is 15 terminal calls per task. The model terminates by calling submit_vulnerable_files with a ranked list, or submit_no_vulnerability_found . The submission itself does not count against the budget.","Output is a ranked list of file paths plus the exploration trace that produced it.","VLoc Bench draws 500 tasks from 290 unique real-world repositories. Sources are public GitHub Security Advisories：https://github.com/advisories across six ecosystems: npm, pip, Maven, Go, Rust, and Composer. It covers 147 unique CWE categories, and 78% of entries carry assigned CVE identifiers.","Ground truth is derived from the security patch. Files modified in the fix are labels, with tests, docs, and configuration excluded.","The pattern in the data is a capability cliff, not a scaling curve.","Antares-3B reaches 0.223 File F1, just under GPT-5.5 (xhigh) at 0.229. Antares-1B reaches 0.209, above GLM-5.2：https://huggingface.co/zai-org at 753B parameters, which scores 0.186. Antares-350M reaches 0.135, above Gemma-4-31B at 0.101 and Gemini 2.5 Flash at 0.102.","Antares-1B also records the highest recall of any evaluated system at 0.224.","Static analysis tools were run under the same evaluation. Semgrep：https://semgrep.dev/ scores 0.086 File F1, CodeQL：https://codeql.github.com/ scores 0.023, and Horusec scores 0.020. Cisco's reading is that rule-based scanners recover some vulnerable files but cannot adaptively inspect repository context.","The untrained Granite 4.0 base checkpoints score 0.001, 0.000, and 0.000 File F1 under the identical protocol. They have tool-calling ability and still produce degenerate output inside an agentic loop.","Supervised fine-tuning does the heavy lifting. It lifts the three scales to 0.108, 0.188, and 0.198. The SFT corpus is 71.5% cybersecurity reasoning, 15.4% code search trajectories, and 13.1% deep research and general reasoning. All reasoning traces come from a single teacher, GPT-OSS-120B, to avoid cross-teacher distribution shift.","GRPO：https://arxiv.org/abs/2402.03300 then adds 11% to 25%, with the largest relative gain at 350M. Rewards are verifiable and computed programmatically from trajectory text, with no learned reward model. Components cover localization quality, submission behavior, tool-use compliance, exploration, and malformed-output penalties.","The variance effect may matter more than the mean. GRPO cuts run-to-run standard deviation by 42% to 65%. One GRPO evaluation run is a more reliable estimate than one SFT run.","There is also a scale-dependent split in learned strategy. After GRPO, the 350M and 1B models use 87% to 89% search commands and submit more files. The 3B model settles at 52% search and 37% read, and submits fewer files at higher precision. The reward never prescribed either policy.","Check out the Models on Hugging Face：https://huggingface.co/collections/fdtn-ai/antares, Benchmark：https://cisco-foundation-ai.github.io/vulnerability-localization-benchmark/, GitHub Repo：https://github.com/cisco-foundation-ai/vulnerability-localization-benchmark and Mentioned Technical Report：https://cisco-foundation-ai.github.io/antares/technical-report.pdf.","Michal Sutter is a data science professional with a Master of Science in Data Science from the University of Padova. With a solid foundation in statistical analysis, machine learning, and data engineering, Michal excels at transforming complex datasets into actionable insights.","Build an Agentic Event Venue Operator [Full Codes]：https://pxllnk.co/twdn5","Thanks! Our team will contact you soon 🙌"],"articleImages":[{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2025/07/a-professional-linkedin-headshot-photogr_0jcmb0R9Sv6nW5XK-zkPHw_uARV5VW1ST6osLNlunoVWg-300x300.png","alt":"","afterParagraph":22,"url":"/media/articles/cmrvqaiiw00csbi85v6yivotz/84e64b03066de40c.webp"},{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2026/07/blog19132-1-8-100x70.png","alt":"Google Releases Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber","afterParagraph":23,"url":"/media/articles/cmrvqaiiw00csbi85v6yivotz/39896896ab8c8014.webp"}],"mediaStatus":"ok","articleBodyZh":["思科基础AI：https://blogs.cisco.com/ai/introducing-antares-the-most-efficient-open-weight-ai-models-for-vulnerability-localization 已发布 Antares，一系列为单一安全任务构建的安全小型语言模型（SLMs）。该任务是漏洞定位。给定漏洞描述和代码库，找到包含漏洞的文件。","两个模型是开源权重，目前已在 Hugging Face 上可用，Antares-350M：https://huggingface.co/fdtn-ai/antares-350m 和 Antares-1B：https://huggingface.co/fdtn-ai/antares-1b。两者均为 Apache 2.0 许可。思科团队还发布了漏洞定位基准测试：https://cisco-foundation-ai.github.io/vulnerability-localization-benchmark/（VLoc Bench），这是一个包含 500 个任务的代理评估，同样在 Apache 2.0 许可下提供。","主要结果并不是新的最先进技术。结果显示，一个 1B 模型达到 0.209 的文件 F1 分数。GPT-5.5 达到 0.229，而一个 753B 的开源权重模型达到 0.186。","软件安全依赖于将外部漏洞知识与内部源代码连接起来。那些知识存在于公共数据库、公告以及常见弱点枚举（https://cwe.mitre.org/）中。代码存在于大型、模块化且依赖丰富的代码库中。","将两者连接起来成本很高。开发者需要搜索不熟悉的代码、遵循命名规范、检查调用路径并比较候选文件。思科的观点是，这个初步筛查步骤是成本集中的地方。","Antares 并不取代应用安全工具链。思科对此非常明确。开发团队仍然需要依赖项扫描、密钥扫描、动态测试、容器检查、威胁建模和专家审查。","Antares 由三个仅解码器的变换模型组成，参数量分别为 350M、1B 和 3B。所有三个模型都从 IBM Granite 4.0 初始化：https://huggingface.co/ibm-granite/granite-4.0-1b 检查点。它们共享一个分词器和架构：分组查询注意力（grouped-query attention）、SwiGLU MLP、RMSNorm、RoPE，以及共享的输入/输出嵌入。","Antares 并未作为独立序列模型进行评估。它在一个受限循环中与三个工具一起运行。","模型仅接收 CWE 分类描述，而不接收其他任何信息。没有咨询文本、没有文件提示、没有严重性细节。然后它会对网络禁用的 Docker 沙箱发出只读终端命令。在进入记录之前，命令输出被截断为 2,000 个字符。","每个任务的预算是 15 次终端调用。模型通过调用 submit_vulnerable_files（带有排名列表）或 submit_no_vulnerability_found 来终止。提交本身不计入预算。","输出是文件路径的排名列表以及产生该列表的探索轨迹。","VLoc Bench 从 290 个独特的真实世界代码库中抽取 500 个任务。来源是公共的 GitHub 安全咨询：https://github.com/advisories，覆盖六个生态系统：npm、pip、Maven、Go、Rust 和 Composer。它涵盖 147 个独特的 CWE 分类，其中 78% 的条目带有已分配的 CVE 标识符。","真实情况来源于安全补丁。修复中修改的文件被标记，测试、文档和配置文件被排除在外。","数据中的模式是能力断崖，而不是扩展曲线。","Antares-3B 达到 0.223 的文件 F1，略低于 GPT-5.5（xhigh）的 0.229。Antares-1B 达到 0.209，高于 GLM-5.2：https://huggingface.co/zai-org 在 753B 参数下的 0.186。Antares-350M 达到 0.135，高于 Gemma-4-31B 的 0.101 和 Gemini 2.5 Flash 的 0.102。","Antares-1B 还记录了任何评估系统中最高的召回率，为 0.224。","静态分析工具在相同评估下运行。Semgrep：https://semgrep.dev/ 得分 0.086 文件 F1，CodeQL：https://codeql.github.com/ 得分 0.023，Horusec 得分 0.020。思科的看法是，基于规则的扫描器可以恢复一些易受攻击的文件，但无法适应性地检查代码库上下文。","未经训练的 Granite 4.0 基础检查点在相同协议下的得分为 0.001、0.000 和 0.000 文件 F1。它们具有工具调用能力，但在具有代理循环的环境中仍会产生退化输出。","监督微调完成了主要工作。它将三个量表提升到0.108、0.188和0.198。SFT语料库中，71.5%用于网络安全推理，15.4%用于代码搜索轨迹，13.1%用于深度研究和通用推理。所有推理轨迹都来自单一教师GPT-OSS-120B，以避免跨教师分布偏移。","GRPO：https://arxiv.org/abs/2402.03300 随后增加了11%到25%，最大相对增益出现在350M。奖励是可验证的，并通过轨迹文本程序化计算，无需学习奖励模型。组件涵盖定位质量、提交行为、工具使用合规性、探索以及格式错误输出惩罚。","方差效应可能比均值更重要。GRPO将运行间标准偏差减少了42%到65%。一次GRPO评估运行比一次SFT运行的估计更可靠。","学习策略也存在依赖规模的差异。GRPO之后，350M和1B模型使用了87%到89%的搜索命令，并提交更多文件。3B模型保持在52%的搜索和37%的阅读，并以更高的精度提交更少文件。奖励从未规定任何策略。","查看Hugging Face上的模型：https://huggingface.co/collections/fdtn-ai/antares，基准测试：https://cisco-foundation-ai.github.io/vulnerability-localization-benchmark/，GitHub仓库：https://github.com/cisco-foundation-ai/vulnerability-localization-benchmark，以及提到的技术报告：https://cisco-foundation-ai.github.io/antares/technical-report.pdf。","Michal Sutter是一名数据科学专业人士，拥有帕多瓦大学数据科学硕士学位。凭借在统计分析、机器学习和数据工程方面的坚实基础，Michal擅长将复杂的数据集转化为可操作的洞见。","构建一个智能事件场馆运营者 [完整代码]：https://pxllnk.co/twdn5","谢谢！我们的团队会尽快联系你 🙌"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于在代码库中定位已知漏洞。 Aioga 将其归入「模型更新」方向，重点关注它对真实使用和行业竞争的影响。","background":"背景分析：模型与研究类动态需要结合能力边界、开放方式、成本、可用性和真实任务表现判断，单项指标领先不等于已经形成稳定采用。","viewpoint":"Aioga 判断：这条动态更适合作为行业观察信号，当前信息足以建立线索，但不足以推导长期结论。","implications":"影响分析：对相关团队而言，短期应先核对来源、可用范围和实际成本，再判断是否值得接入或跟进。","nextStep":"后续观察：继续观察官方文档、实际可用性、价格变化、开发者反馈和竞品回应。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-07-23T06:49:19.042Z","sourceHash":"07693952338a7cf2","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["模型更新","MarkTechPost（RSS）"],"translations":{"zh-CN":{"title":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于漏洞定位","summary":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于在代码库中定位已知漏洞。Antares-1B 在 VLoc Bench 上达到 0.209 File F1，超过 753B 参数的 GLM-5.2（0.186）。Antares-350M 和 Antares-1B 以 Apache 2.0 许可证开源。","category":"模型更新","source":"marktechpost.com","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于漏洞定位 - Aioga AI资讯","description":"Cisco Foundation AI 发布 Antares 系列安全小语言模型，专用于在代码库中定位已知漏洞。Antares-1B 在 VLoc Bench 上达到 0.209 File F1，超过 753B 参数的 GLM-5.2（0.186）。Antares-350M 和 Antares-1B 以 Apache 2.0 许可证开源。","url":"https://www.aioga.com/news/cmrvqaiiw00csbi85v6yivotz/"},"en":{"title":"Cisco Foundation AI launches the Antares series of small security language models, specifically for vulnerability detection","summary":"Cisco Foundation AI released the Antares series of small security language models, specifically designed to locate known vulnerabilities in codebases. Antares-1B achieved a File F1 of 0.209 on VLoc Bench, surpassing GLM-5.2 with 753B parameters (0.186). Antares-350M and Antares-1B are open-sourced under the Apache 2.0 license.","category":"Models","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI launches the Antares series of small security language models, specifically for vulnerability detection - Aioga AI News","description":"Cisco Foundation AI released the Antares series of small security language models, specifically designed to locate known vulnerabilities in codebases. Antares-1B achieved a File F1...","url":"https://www.aioga.com/en/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:22:53.622Z"},"ja":{"title":"Cisco Foundation AIはAntaresシリーズのセキュリティ用小型言語モデルを発表、脆弱性の特定専用","summary":"Cisco Foundation AI は Antares シリーズのセキュリティ用小型言語モデルを発表し、コードベース内の既知の脆弱性の特定に特化しています。Antares-1B は VLoc Bench で 0.209 File F1 を達成し、753B パラメータの GLM-5.2（0.186）を上回りました。Antares-350M と Antares-1B は Apache 2.0 ライセンスでオープンソースとして提供されます。","category":"モデル更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AIはAntaresシリーズのセキュリティ用小型言語モデルを発表、脆弱性の特定専用 - Aioga AIニュース","description":"Cisco Foundation AI は Antares シリーズのセキュリティ用小型言語モデルを発表し、コードベース内の既知の脆弱性の特定に特化しています。Antares-1B は VLoc Bench で 0.209 File F1 を達成し、753B パラメータの GLM-5.2（0.186）を上回りました。Antares-350M と Antare...","url":"https://www.aioga.com/ja/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:23:03.253Z"},"ko":{"title":"Cisco Foundation AI가 Antares 시리즈 보안 소형 언어 모델을 출시했으며, 취약점 위치 지정 전용입니다","summary":"Cisco Foundation AI가 Antares 시리즈 보안 소형 언어 모델을 출시했으며, 이는 코드베이스에서 알려진 취약점을 찾는 데 전용으로 사용됩니다. Antares-1B는 VLoc Bench에서 0.209 File F1을 기록하여 753B 파라미터의 GLM-5.2(0.186)를 능가했습니다. Antares-350M과 Antares-1B는 Apache 2.0 라이선스로 오픈소스로 제공됩니다.","category":"모델 업데이트","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI가 Antares 시리즈 보안 소형 언어 모델을 출시했으며, 취약점 위치 지정 전용입니다 - Aioga AI 뉴스","description":"Cisco Foundation AI가 Antares 시리즈 보안 소형 언어 모델을 출시했으며, 이는 코드베이스에서 알려진 취약점을 찾는 데 전용으로 사용됩니다. Antares-1B는 VLoc Bench에서 0.209 File F1을 기록하여 753B 파라미터의 GLM-5.2(0.186)를 능가했습니다. Antares-35...","url":"https://www.aioga.com/ko/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:23:44.751Z"},"es":{"title":"Cisco Foundation AI lanza la serie Antares de pequeños modelos de lenguaje de seguridad, especialmente para la localización de vulnerabilidades","summary":"Cisco Foundation AI lanzó la serie de pequeños modelos de lenguaje de seguridad Antares, diseñada específicamente para localizar vulnerabilidades conocidas en repositorios de código. Antares-1B alcanzó un F1 de archivo de 0,209 en VLoc Bench, superando a GLM-5.2 de 753B parámetros (0,186). Antares-350M y Antares-1B se publican como código abierto bajo la licencia Apache 2.0.","category":"Modelos","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI lanza la serie Antares de pequeños modelos de lenguaje de seguridad, especialmente para la localización de vulnerabilidades - Aioga Noticias de IA","description":"Cisco Foundation AI lanzó la serie de pequeños modelos de lenguaje de seguridad Antares, diseñada específicamente para localizar vulnerabilidades conocidas en repositorios de códig...","url":"https://www.aioga.com/es/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:23:41.879Z"},"fr":{"title":"Cisco Foundation AI lance la série de petits modèles de langage sécurisés Antares, spécialement conçue pour la localisation des vulnérabilités","summary":"Cisco Foundation AI a publié la série de petits modèles de langage de sécurité Antares, spécialement conçus pour localiser les vulnérabilités connues dans les bases de code. Antares-1B atteint 0,209 File F1 sur VLoc Bench, surpassant GLM-5.2 de 753 milliards de paramètres (0,186). Antares-350M et Antares-1B sont open source sous licence Apache 2.0.","category":"Modèles","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI lance la série de petits modèles de langage sécurisés Antares, spécialement conçue pour la localisation des vulnérabilités - Aioga Actualités IA","description":"Cisco Foundation AI a publié la série de petits modèles de langage de sécurité Antares, spécialement conçus pour localiser les vulnérabilités connues dans les bases de code. Antare...","url":"https://www.aioga.com/fr/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:24:23.821Z"},"de":{"title":"Cisco Foundation AI veröffentlicht die Antares-Serie sicherer kleiner Sprachmodelle, speziell für die Schwachstellenlokalisierung","summary":"Cisco Foundation AI veröffentlicht die Antares-Serie von kleinen Sicherheits-Sprachmodellen, die speziell dafür entwickelt wurden, bekannte Schwachstellen in Codebasen zu lokalisieren. Antares-1B erreicht auf dem VLoc Bench eine File F1 von 0,209 und übertrifft damit das 753B-Parameter-Modell GLM-5.2 (0,186). Antares-350M und Antares-1B werden unter der Apache 2.0-Lizenz open-source veröffentlicht.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI veröffentlicht die Antares-Serie sicherer kleiner Sprachmodelle, speziell für die Schwachstellenlokalisierung - Aioga KI-News","description":"Cisco Foundation AI veröffentlicht die Antares-Serie von kleinen Sicherheits-Sprachmodellen, die speziell dafür entwickelt wurden, bekannte Schwachstellen in Codebasen zu lokalisie...","url":"https://www.aioga.com/de/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:24:24.379Z"},"pt-BR":{"title":"A Cisco Foundation AI lançou a série de pequenos modelos de linguagem de segurança Antares, especialmente para localização de vulnerabilidades","summary":"A Cisco Foundation AI lançou a série Antares de pequenos modelos de linguagem focados em segurança, especialmente projetados para localizar vulnerabilidades conhecidas em repositórios de código. O Antares-1B atingiu 0,209 de File F1 no VLoc Bench, superando o GLM-5.2 de 753B parâmetros (0,186). O Antares-350M e o Antares-1B são de código aberto sob a licença Apache 2.0.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"A Cisco Foundation AI lançou a série de pequenos modelos de linguagem de segurança Antares, especialmente para localização de vulnerabilidades - Aioga Notícias de IA","description":"A Cisco Foundation AI lançou a série Antares de pequenos modelos de linguagem focados em segurança, especialmente projetados para localizar vulnerabilidades conhecidas em repositór...","url":"https://www.aioga.com/pt-BR/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:25:03.499Z"},"ru":{"title":"Фонд Cisco AI выпустил серию безопасных малых языковых моделей Antares, специально предназначенных для обнаружения уязвимостей","summary":"Cisco Foundation AI выпустила серию моделей безопасных малых языковых моделей Antares, специально предназначенных для обнаружения известных уязвимостей в кодовой базе. Antares-1B достигла 0.209 File F1 на VLoc Bench, превзойдя GLM-5.2 с 753B параметрами (0.186). Antares-350M и Antares-1B выпущены с открытым исходным кодом по лицензии Apache 2.0.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Фонд Cisco AI выпустил серию безопасных малых языковых моделей Antares, специально предназначенных для обнаружения уязвимостей - Aioga Новости ИИ","description":"Cisco Foundation AI выпустила серию моделей безопасных малых языковых моделей Antares, специально предназначенных для обнаружения известных уязвимостей в кодовой базе. Antares-1B д...","url":"https://www.aioga.com/ru/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:25:07.178Z"},"ar":{"title":"أصدرت Cisco Foundation AI سلسلة Antares من النماذج اللغوية الصغيرة للأمن، المخصصة لتحديد الثغرات","summary":"أصدرت شركة Cisco Foundation AI سلسلة نماذج اللغة الصغيرة Antares للأمان، والمخصصة لتحديد الثغرات المعروفة في مستودعات الكود. سجل نموذج Antares-1B نسبة F1 للملفات 0.209 على VLoc Bench، متفوقًا على GLM-5.2 الذي يحتوي على 753 مليار معامل بنسبة 0.186. تم إصدار كل من Antares-350M و Antares-1B كمصدر مفتوح برخصة Apache 2.0.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"أصدرت Cisco Foundation AI سلسلة Antares من النماذج اللغوية الصغيرة للأمن، المخصصة لتحديد الثغرات - Aioga أخبار الذكاء الاصطناعي","description":"أصدرت شركة Cisco Foundation AI سلسلة نماذج اللغة الصغيرة Antares للأمان، والمخصصة لتحديد الثغرات المعروفة في مستودعات الكود. سجل نموذج Antares-1B نسبة F1 للملفات 0.209 على VLoc Ben...","url":"https://www.aioga.com/ar/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:25:52.035Z"},"hi":{"title":"सिस्को फाउंडेशन एआई ने अंटार्स श्रृंखला का सुरक्षा लघु भाषा मॉडल लॉन्च किया, जो विशेष रूप से कमजोरियों का पता लगाने के लिए है","summary":"Cisco Foundation AI ने Antares श्रृंखला की सुरक्षा लघु भाषा मॉडल जारी की, जो विशेष रूप से कोडबेस में ज्ञात कमजोरियों का पता लगाने के लिए डिज़ाइन की गई हैं। Antares-1B ने VLoc Bench पर 0.209 फ़ाइल F1 का स्कोर प्राप्त किया, जो 753B पैरामीटर वाले GLM-5.2 (0.186) से अधिक है। Antares-350M और Antares-1B को Apache 2.0 लाइसेंस के तहत ओपन सोर्स किया गया है।","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"सिस्को फाउंडेशन एआई ने अंटार्स श्रृंखला का सुरक्षा लघु भाषा मॉडल लॉन्च किया, जो विशेष रूप से कमजोरियों का पता लगाने के लिए है - Aioga AI समाचार","description":"Cisco Foundation AI ने Antares श्रृंखला की सुरक्षा लघु भाषा मॉडल जारी की, जो विशेष रूप से कोडबेस में ज्ञात कमजोरियों का पता लगाने के लिए डिज़ाइन की गई हैं। Antares-1B ने VLoc Bench...","url":"https://www.aioga.com/hi/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:25:51.924Z"},"it":{"title":"Cisco Foundation AI ha rilasciato la serie Antares di piccoli modelli linguistici per la sicurezza, progettati specificamente per il rilevamento delle vulnerabilità","summary":"Cisco Foundation AI ha rilasciato la serie di piccoli modelli di linguaggio sicuri Antares, progettata specificamente per individuare vulnerabilità note nei repository di codice. Antares-1B ha raggiunto 0,209 File F1 su VLoc Bench, superando GLM-5.2 con 753 miliardi di parametri (0,186). Antares-350M e Antares-1B sono open source con licenza Apache 2.0.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI ha rilasciato la serie Antares di piccoli modelli linguistici per la sicurezza, progettati specificamente per il rilevamento delle vulnerabilità - Aioga Notizie IA","description":"Cisco Foundation AI ha rilasciato la serie di piccoli modelli di linguaggio sicuri Antares, progettata specificamente per individuare vulnerabilità note nei repository di codice. A...","url":"https://www.aioga.com/it/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:26:31.068Z"},"nl":{"title":"Cisco Foundation AI lanceert de Antares-serie veilige kleine taalmodellen, speciaal voor het lokaliseren van kwetsbaarheden","summary":"Cisco Foundation AI heeft de Antares-serie beveiligingsgerichte kleine taalmodellen uitgebracht, speciaal ontworpen om bekende kwetsbaarheden in codebases te lokaliseren. Antares-1B behaalde een File F1-score van 0,209 op VLoc Bench, beter dan GLM-5.2 met 753 miljard parameters (0,186). Antares-350M en Antares-1B zijn open source onder de Apache 2.0-licentie.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI lanceert de Antares-serie veilige kleine taalmodellen, speciaal voor het lokaliseren van kwetsbaarheden - Aioga AI-nieuws","description":"Cisco Foundation AI heeft de Antares-serie beveiligingsgerichte kleine taalmodellen uitgebracht, speciaal ontworpen om bekende kwetsbaarheden in codebases te lokaliseren. Antares-1...","url":"https://www.aioga.com/nl/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:26:31.537Z"},"tr":{"title":"Cisco Foundation AI, Antares serisi güvenlik küçük dil modellerini yayınladı, özel olarak güvenlik açıklarını tespit etmek için","summary":"Cisco Foundation AI, Antares serisi güvenlik küçük dil modellerini yayınladı, bu modeller özellikle kod tabanında bilinen güvenlik açıklarını bulmak için tasarlandı. Antares-1B, VLoc Bench üzerinde 0.209 File F1 değerine ulaştı ve 753B parametreli GLM-5.2'nin (0.186) üzerinde performans gösterdi. Antares-350M ve Antares-1B, Apache 2.0 lisansı ile açık kaynak olarak sunulmaktadır.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI, Antares serisi güvenlik küçük dil modellerini yayınladı, özel olarak güvenlik açıklarını tespit etmek için - Aioga AI Haberleri","description":"Cisco Foundation AI, Antares serisi güvenlik küçük dil modellerini yayınladı, bu modeller özellikle kod tabanında bilinen güvenlik açıklarını bulmak için tasarlandı. Antares-1B, VL...","url":"https://www.aioga.com/tr/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:27:13.496Z"},"vi":{"title":"Cisco Foundation AI ra mắt loạt mô hình ngôn ngữ nhỏ Antares chuyên dụng cho định vị lỗ hổng","summary":"Cisco Foundation AI ra mắt loạt mô hình ngôn ngữ nhỏ Antares chuyên dụng cho việc định vị các lỗ hổng đã biết trong kho mã. Antares-1B đạt 0,209 File F1 trên VLoc Bench, vượt qua GLM-5.2 với 753 tỷ tham số (0,186). Antares-350M và Antares-1B được phát hành mã nguồn mở theo giấy phép Apache 2.0.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI ra mắt loạt mô hình ngôn ngữ nhỏ Antares chuyên dụng cho định vị lỗ hổng - Tin tức AI Aioga","description":"Cisco Foundation AI ra mắt loạt mô hình ngôn ngữ nhỏ Antares chuyên dụng cho việc định vị các lỗ hổng đã biết trong kho mã. Antares-1B đạt 0,209 File F1 trên VLoc Bench, vượt qua G...","url":"https://www.aioga.com/vi/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:27:11.291Z"},"id":{"title":"Cisco Foundation AI merilis seri model bahasa kecil Antares yang khusus untuk penentuan lokasi kerentanan","summary":"Cisco Foundation AI merilis model bahasa kecil keamanan seri Antares, khusus digunakan untuk menemukan kerentanan yang diketahui dalam repositori kode. Antares-1B mencapai 0,209 File F1 di VLoc Bench, melampaui GLM-5.2 dengan 753 miliar parameter (0,186). Antares-350M dan Antares-1B dirilis sebagai open source dengan lisensi Apache 2.0.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI merilis seri model bahasa kecil Antares yang khusus untuk penentuan lokasi kerentanan - Berita AI Aioga","description":"Cisco Foundation AI merilis model bahasa kecil keamanan seri Antares, khusus digunakan untuk menemukan kerentanan yang diketahui dalam repositori kode. Antares-1B mencapai 0,209 Fi...","url":"https://www.aioga.com/id/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:27:52.471Z"},"th":{"title":"Cisco Foundation AI เปิดตัวโมเดลภาษาเล็กด้านความปลอดภัยชุด Antares โดยเฉพาะสำหรับการระบุช่องโหว่","summary":"Cisco Foundation AI เปิดตัวโมเดลภาษาเล็กด้านความปลอดภัยชุด Antares ซึ่งออกแบบมาโดยเฉพาะเพื่อระบุตำแหน่งช่องโหว่ที่ทราบในฐานรหัส Antares-1B ทำคะแนน 0.209 File F1 ใน VLoc Bench สูงกว่า GLM-5.2 ที่มีพารามิเตอร์ 753B (0.186) Antares-350M และ Antares-1B เปิดซอร์สภายใต้สัญญาอนุญาต Apache 2.0","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Cisco Foundation AI เปิดตัวโมเดลภาษาเล็กด้านความปลอดภัยชุด Antares โดยเฉพาะสำหรับการระบุช่องโหว่ - ข่าว AI Aioga","description":"Cisco Foundation AI เปิดตัวโมเดลภาษาเล็กด้านความปลอดภัยชุด Antares ซึ่งออกแบบมาโดยเฉพาะเพื่อระบุตำแหน่งช่องโหว่ที่ทราบในฐานรหัส Antares-1B ทำคะแนน 0.209 File F1 ใน VLoc Bench สูงกว...","url":"https://www.aioga.com/th/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:27:59.134Z"},"pl":{"title":"Fundacja Cisco AI wydaje serię małych modeli językowych Antares do bezpieczeństwa, specjalnie do lokalizowania luk","summary":"Cisco Foundation AI ogłosiło serię modeli językowych Antares do bezpieczeństwa, specjalnie przeznaczonych do lokalizowania znanych luk w bibliotekach kodu. Antares-1B osiągnął 0,209 File F1 w VLoc Bench, przewyższając GLM-5.2 z 753 miliardami parametrów (0,186). Antares-350M i Antares-1B są otwartoźródłowe na licencji Apache 2.0.","category":"模型更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Fundacja Cisco AI wydaje serię małych modeli językowych Antares do bezpieczeństwa, specjalnie do lokalizowania luk - Aioga Wiadomości AI","description":"Cisco Foundation AI ogłosiło serię modeli językowych Antares do bezpieczeństwa, specjalnie przeznaczonych do lokalizowania znanych luk w bibliotekach kodu. Antares-1B osiągnął 0,20...","url":"https://www.aioga.com/pl/news/cmrvqaiiw00csbi85v6yivotz/","contentTranslated":true,"sourceHash":"57e2c29097ba6d30","translatedAt":"2026-07-22T23:28:47.053Z"}}}}