{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-09-22T18:00:52.144Z","headline":"Sierra 开源 hyper-τ-bench：评估能构建智能体的智能体","description":"Sierra 今日开源 hyper-τ-bench，这是一个新的长时程智能体评测基准，用于衡量模型不仅能作为智能体行动、还能构建智能体的能力。该基准聚焦于\"构建智能体的智能体\"这一任务场景，为长时程自主智能体开发提供评测标准。","url":"https://www.aioga.com/news/cmtt9xrai020orofps61otpfi/","mainEntityOfPage":"https://www.aioga.com/news/cmtt9xrai020orofps61otpfi/","datePublished":"2026-09-08T20:25:11.000Z","dateModified":"2026-09-08T20:25:11.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://sierra.ai/blog/hyper-t-bench-evaluating-agents-that-build-agents","https://aihot.news/items/cmtt9xrai020orofps61otpfi"],"canonicalUrl":"https://www.aioga.com/news/cmtt9xrai020orofps61otpfi/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：Sierra 今日开源 hyper-τ-bench，这是一个新的长时程智能体评测基准，用于衡量模型不仅能作为智能体行动、还能构建智能体的能力。 Aioga 将其归入「行业动态」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/cmtt9xrai020orofps61otpfi/","dateCreated":"2026-09-08T20:25:11.000Z","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":"sierra.ai source article","url":"https://sierra.ai/blog/hyper-t-bench-evaluating-agents-that-build-agents","datePublished":"2026-09-08T20:25:11.000Z","provider":{"@type":"Organization","name":"sierra.ai","url":"https://sierra.ai/blog/hyper-t-bench-evaluating-agents-that-build-agents"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.news/items/cmtt9xrai020orofps61otpfi","datePublished":"2026-09-08T20:25:11.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.news/items/cmtt9xrai020orofps61otpfi"}}],"aggregationSource":"Sierra：Blog（RSS）","originalPublisher":{"name":"sierra.ai","url":"https://sierra.ai/blog/hyper-t-bench-evaluating-agents-that-build-agents"},"geoDeepAnswer":null,"article":{"id":"cmtt9xrai020orofps61otpfi","slug":"cmtt9xrai020orofps61otpfi","url":"https://www.aioga.com/news/cmtt9xrai020orofps61otpfi/","title":"Sierra 开源 hyper-τ-bench：评估能构建智能体的智能体","title_en":"","summary":"Sierra 今日开源 hyper-τ-bench，这是一个新的长时程智能体评测基准，用于衡量模型不仅能作为智能体行动、还能构建智能体的能力。该基准聚焦于\"构建智能体的智能体\"这一任务场景，为长时程自主智能体开发提供评测标准。","source":"Sierra：Blog（RSS）","sourceUrl":"https://sierra.ai/blog/hyper-t-bench-evaluating-agents-that-build-agents","aiHotUrl":"https://aihot.news/items/cmtt9xrai020orofps61otpfi","publishedAt":"2026-09-08T20:25:11.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["Get notified about new product features, customer updates, and more.","We built 𝜏-bench in 2024：https://sierra.ai/blog/benchmarking-ai-agents to answer a question that felt novel at the time: Can a model act as a reliable customer service agent? That’s table stakes now. The harder question is, who’s building the agent in the first place? Increasingly, it’s the models themselves.","We’ve partnered closely with some of the world’s leading companies to launch their agents. In practice, the work is less like implementing a spec, and more like doing research. Requirements are scattered across handbooks, support, spreadsheets, and the minds of your best frontline reps — so you form a hypothesis, dig up evidence, and build and test to identify which levers actually move performance.","Today we’re open-sourcing hyper-𝜏-bench (published as 𝜏^𝜏-bench), a new long horizon agent evaluation that measures how well models can not only act as an agent, but construct one.","Hyper-𝜏-bench drops a developer agent into a sandboxed workspace with the records of a simulated business, plus a simulated client that it can message at any time. From there, the developer agent does the job end-to-end — it recovers the spec from the evidence, designs the architecture, and turns the business’s actions into tools — until it has a working customer-service agent. The client’s REST API may be subtly defective, so part of the job is figuring out whether a bug is in the spec or in the code. The finished agent has to serve from a fixed menu of models, within a cost budget per conversation. Once it’s handed off, we deploy it against simulated production traffic using fully verifiable 𝜏-bench-style tests the developer never saw while building.","Working alone, our best configuration — Claude Opus 5 (max reasoning) running in Claude Code — passes just 23.9% of the held-out evaluation tasks. Paired with an engineer with deep context, the same class of model reaches 82.2% on the same tasks.","We read through developer trajectories to see where their builds lost ground. Five patterns stood out:","Hyper-𝜏-bench sits alongside benchmarks like MLE-bench and RE-Bench, which measure research capability: designing experiments, weighing tradeoffs, and iterating toward a better system. Building an agent demands all of that — and adds a few problems of its own. The spec has to be recovered from documents and people. And because the system being built is an AI itself, the only way to know if a design works is to run it and read what it says to real users, who the developer never sees while building.","𝜏-bench asked whether models could be good agents. Hyper-𝜏-bench asks whether they can build them. As agents take on more of that work themselves, we’ll keep using hyper-𝜏-bench to track how well they’re doing it.","Paper：https://arxiv.org/abs/2609.04611 | Codebase：https://github.com/sierra-research/hyper-tau-bench | Leaderboard：https://sierra-research.github.io/hyper-tau-bench/","Find out how Sierra can help you deliver better outcomes with AI.","𝜏³-Bench is here. We've expanded agent evaluation to two new frontiers: knowledge retrieval and voice.","𝜏²-bench challenges AI agents not just to reason and act, but to coordinate, guide, and assist a user in achieving a shared objective. This leap from solo operation to co-ownership of a task pushes agents into a much more demanding space. And, critically, it reflects the kinds of tasks AI agents are increasingly being asked to perform in the real world.","We explore how Sierra’s 𝜏-bench is shaping the development and evaluation of agents—from academic research to industry applications and next-generation development."],"articleImages":[{"sourceUrl":"https://sierra.ai/-/cdn/image?src=https%3A%2F%2Fcdn.sanity.io%2Fimages%2Fca4jck6w%2Fproduction%2Fe4acb9f1db7a8b9935cfc1408cc79956e2f14f33-512x512.jpg&width=3840&quality=75","alt":"Headshot of Ben Shi","afterParagraph":0,"url":"/media/articles/cmtt9xrai020orofps61otpfi/dda95b55b2de5f7c.avif"},{"sourceUrl":"https://sierra.ai/-/cdn/image?src=https%3A%2F%2Fcdn.sanity.io%2Fimages%2Fca4jck6w%2Fproduction%2F0ea7f00af0972dbd30a3fb3f3249a5e371316660-478x512.jpg&width=3840&quality=75","alt":"Headshot of Keshav Dhandhania","afterParagraph":0,"url":"/media/articles/cmtt9xrai020orofps61otpfi/66f1bbe2cc66a32f.avif"},{"sourceUrl":"https://sierra.ai/-/cdn/image?src=https%3A%2F%2Fcdn.sanity.io%2Fimages%2Fca4jck6w%2Fproduction%2F3651521e8f0ff7f560f0e16a8afe5756d9360c33-1200x800.png&width=3840&quality=75","alt":"The text \"τ³-Bench\" on a blurry, grainy background of green and brown.","afterParagraph":10,"url":"/media/articles/cmtt9xrai020orofps61otpfi/58f848cca5f31e65.avif"},{"sourceUrl":"https://sierra.ai/-/cdn/image?src=https%3A%2F%2Fcdn.sanity.io%2Fimages%2Fca4jck6w%2Fproduction%2Fc083407f50cd3e73b9d1fef72cfc5682494b42ee-1200x800.png&width=3840&quality=75","alt":"𝜏²-bench","afterParagraph":11,"url":"/media/articles/cmtt9xrai020orofps61otpfi/04d3b774b833bbca.avif"},{"sourceUrl":"https://sierra.ai/-/cdn/image?src=https%3A%2F%2Fcdn.sanity.io%2Fimages%2Fca4jck6w%2Fproduction%2F7037514b171647ae75039fc236fe11d4d2237cdc-2160x2160.png&width=3840&quality=75","alt":"𝜏-bench","afterParagraph":12,"url":"/media/articles/cmtt9xrai020orofps61otpfi/bf13f51604c7f167.webp"}],"mediaStatus":"ok","articleBodyZh":["获取有关新产品功能、客户更新等的通知。","我们在 2024 年构建了𝜏-bench：https://sierra.ai/blog/benchmarking-ai-agents 来回答一个当时感觉很新颖的问题：模型能否充当可靠的客户服务代理？这就是现在的赌注。更难的问题是，谁首先构建代理？越来越多的是模型本身。","我们与一些世界领先的公司紧密合作，以推出他们的代理。在实践中，这项工作不像执行规范，更像是在做研究。需求分散在手册、支持文档、电子表格以及你最优秀的前线员工的脑海中——所以你提出假设，寻找证据，并构建和测试，以确定哪些杠杆实际上能提升性能。","今天我们开源了hyper-𝜏-bench（以𝜏^𝜏-bench发布），这是一个新的长期视角代理评估，它衡量模型不仅能作为代理工作，还能构建代理的能力。","Hyper-𝜏-bench将开发者代理放入一个沙箱工作区，里面有模拟业务的记录，以及一个随时可以消息的模拟客户。然后，开发者代理从头到尾完成工作——从证据中恢复规范，设计架构，并将业务操作转化为工具——直到拥有一个可用的客服代理。客户的REST API可能存在微小缺陷，所以部分工作是判断是规范的问题还是代码的问题。完成的代理必须在每次对话的成本预算内，从固定的模型菜单中提供服务。一旦交付，我们会使用开发者构建过程中未见过的、完全可验证的𝜏-bench式测试，将其部署到模拟生产流量中。","单独工作时，我们的最佳配置——运行Claude Code的Claude Opus 5（最大推理）——仅通过23.9%的保留评估任务。与具有深厚背景知识的工程师配对时，相同类别的模型在相同任务中可达到82.2%。","我们阅读了开发者的轨迹，以查看他们的构建在哪些地方失分。有五种模式显著突出：","Hyper-𝜏-bench 与 MLE-bench 和 RE-Bench 等评测基准并列，这些基准衡量研究能力：设计实验、权衡利弊以及迭代以改进系统。构建一个智能体要求具备所有这些能力——并且额外增加一些自己的问题。规范必须从文档和人员中获取。而且因为正在构建的系统本身就是一个 AI，唯一能够判断设计是否有效的方法就是运行它，并读取它对真实用户的输出，而开发者在构建过程中从未见过这些用户。","𝜏-bench 关注模型是否可以成为优秀的智能体。Hyper-𝜏-bench 则关注它们是否能构建智能体。随着智能体越来越多地承担这些工作，我们将继续使用 hyper-𝜏-bench 来追踪它们的表现。","论文：https://arxiv.org/abs/2609.04611 | 代码库：https://github.com/sierra-research/hyper-tau-bench | 排行榜：https://sierra-research.github.io/hyper-tau-bench/","了解 Sierra 如何帮助你用 AI 实现更好的成果。","𝜏³-Bench 已经到来。我们将智能体评估扩展到了两个新前沿：知识检索和语音。","𝜏²-bench 不仅要求 AI 智能体进行推理和行动，还要求它们协调、指导并协助用户实现共同目标。从单独操作到任务共管，这一跃迁将智能体推向了更具挑战性的领域。而且，关键是，它反映了现实世界中 AI 智能体越来越多被要求执行的任务类型。","我们探讨 Sierra 的 𝜏-bench 如何影响智能体的开发和评估——从学术研究到行业应用及下一代开发。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：Sierra 今日开源 hyper-τ-bench，这是一个新的长时程智能体评测基准，用于衡量模型不仅能作为智能体行动、还能构建智能体的能力。 Aioga 将其归入「行业动态」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":"2026-09-22T18:20:58.004Z","sourceHash":"4bde354de7841fef","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["行业动态","Sierra：Blog（RSS）"],"translations":{"zh-CN":{"title":"Sierra 开源 hyper-τ-bench：评估能构建智能体的智能体","summary":"Sierra 今日开源 hyper-τ-bench，这是一个新的长时程智能体评测基准，用于衡量模型不仅能作为智能体行动、还能构建智能体的能力。该基准聚焦于\"构建智能体的智能体\"这一任务场景，为长时程自主智能体开发提供评测标准。","category":"行业动态","source":"sierra.ai","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra 开源 hyper-τ-bench：评估能构建智能体的智能体 - Aioga AI资讯","description":"Sierra 今日开源 hyper-τ-bench，这是一个新的长时程智能体评测基准，用于衡量模型不仅能作为智能体行动、还能构建智能体的能力。该基准聚焦于\"构建智能体的智能体\"这一任务场景，为长时程自主智能体开发提供评测标准。","url":"https://www.aioga.com/news/cmtt9xrai020orofps61otpfi/","articleBody":["获取有关新产品功能、客户更新等的通知。","我们在 2024 年构建了𝜏-bench：https://sierra.ai/blog/benchmarking-ai-agents 来回答一个当时感觉很新颖的问题：模型能否充当可靠的客户服务代理？这就是现在的赌注。更难的问题是，谁首先构建代理？越来越多的是模型本身。","我们与一些世界领先的公司紧密合作，以推出他们的代理。在实践中，这项工作不像执行规范，更像是在做研究。需求分散在手册、支持文档、电子表格以及你最优秀的前线员工的脑海中——所以你提出假设，寻找证据，并构建和测试，以确定哪些杠杆实际上能提升性能。","今天我们开源了hyper-𝜏-bench（以𝜏^𝜏-bench发布），这是一个新的长期视角代理评估，它衡量模型不仅能作为代理工作，还能构建代理的能力。","Hyper-𝜏-bench将开发者代理放入一个沙箱工作区，里面有模拟业务的记录，以及一个随时可以消息的模拟客户。然后，开发者代理从头到尾完成工作——从证据中恢复规范，设计架构，并将业务操作转化为工具——直到拥有一个可用的客服代理。客户的REST API可能存在微小缺陷，所以部分工作是判断是规范的问题还是代码的问题。完成的代理必须在每次对话的成本预算内，从固定的模型菜单中提供服务。一旦交付，我们会使用开发者构建过程中未见过的、完全可验证的𝜏-bench式测试，将其部署到模拟生产流量中。","单独工作时，我们的最佳配置——运行Claude Code的Claude Opus 5（最大推理）——仅通过23.9%的保留评估任务。与具有深厚背景知识的工程师配对时，相同类别的模型在相同任务中可达到82.2%。","我们阅读了开发者的轨迹，以查看他们的构建在哪些地方失分。有五种模式显著突出：","Hyper-𝜏-bench 与 MLE-bench 和 RE-Bench 等评测基准并列，这些基准衡量研究能力：设计实验、权衡利弊以及迭代以改进系统。构建一个智能体要求具备所有这些能力——并且额外增加一些自己的问题。规范必须从文档和人员中获取。而且因为正在构建的系统本身就是一个 AI，唯一能够判断设计是否有效的方法就是运行它，并读取它对真实用户的输出，而开发者在构建过程中从未见过这些用户。","𝜏-bench 关注模型是否可以成为优秀的智能体。Hyper-𝜏-bench 则关注它们是否能构建智能体。随着智能体越来越多地承担这些工作，我们将继续使用 hyper-𝜏-bench 来追踪它们的表现。","论文：https://arxiv.org/abs/2609.04611 | 代码库：https://github.com/sierra-research/hyper-tau-bench | 排行榜：https://sierra-research.github.io/hyper-tau-bench/","了解 Sierra 如何帮助你用 AI 实现更好的成果。","𝜏³-Bench 已经到来。我们将智能体评估扩展到了两个新前沿：知识检索和语音。","𝜏²-bench 不仅要求 AI 智能体进行推理和行动，还要求它们协调、指导并协助用户实现共同目标。从单独操作到任务共管，这一跃迁将智能体推向了更具挑战性的领域。而且，关键是，它反映了现实世界中 AI 智能体越来越多被要求执行的任务类型。","我们探讨 Sierra 的 𝜏-bench 如何影响智能体的开发和评估——从学术研究到行业应用及下一代开发。"]},"en":{"title":"Sierra open-source hyper-τ-bench: Evaluates agents capable of building agents","summary":"Sierra today open-sourced hyper-τ-bench, a new long-duration agent evaluation benchmark designed to measure a model's ability not only to act as an agent but also to build agents. This benchmark focuses on the task scenario of \"building agents,\" providing evaluation standards for long-duration autonomous agent development.","category":"Industry","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: Evaluates agents capable of building agents - Aioga AI News","description":"Sierra today open-sourced hyper-τ-bench, a new long-duration agent evaluation benchmark designed to measure a model's ability not only to act as an agent but also to build agents....","url":"https://www.aioga.com/en/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:25.980Z"},"ja":{"title":"Sierraオープンソースのhyper-τ-bench:エージェントをビルドできるエージェントを評価する","summary":"Sierraは本日、モデルがエージェントとして機能するだけでなく、エージェントを構築する能力を測定するために設計された新しい長期エージェント評価ベンチマークであるHyper-τ-benchをオープンソース化しました。このベンチマークは「エージェント構築」のタスクシナリオに焦点を当て、長期自律型エージェント開発の評価基準を提供します。","category":"業界動向","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierraオープンソースのhyper-τ-bench:エージェントをビルドできるエージェントを評価する - Aioga AIニュース","description":"Sierraは本日、モデルがエージェントとして機能するだけでなく、エージェントを構築する能力を測定するために設計された新しい長期エージェント評価ベンチマークであるHyper-τ-benchをオープンソース化しました。このベンチマークは「エージェント構築」のタスクシナリオに焦点を当て、長期自律型エージェント開発の評価基準を提供します。","url":"https://www.aioga.com/ja/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:25.611Z"},"ko":{"title":"Sierra 오픈 소스 hyper-τ-bench: 에이전트를 구축할 수 있는 에이전트를 평가합니다","summary":"시에라는 오늘 모델이 에이전트로서 행동할 뿐만 아니라 에이전트를 구축하는 능력을 측정하기 위해 설계된 새로운 장기 에이전트 평가 벤치마크인 hyper-τ-bench를 오픈 소스로 출시했습니다. 이 벤치마크는 \"에이전트 구축\" 작업 시나리오에 초점을 맞추어 장기 자율 에이전트 개발을 위한 평가 기준을 제공합니다.","category":"업계 동향","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra 오픈 소스 hyper-τ-bench: 에이전트를 구축할 수 있는 에이전트를 평가합니다 - Aioga AI 뉴스","description":"시에라는 오늘 모델이 에이전트로서 행동할 뿐만 아니라 에이전트를 구축하는 능력을 측정하기 위해 설계된 새로운 장기 에이전트 평가 벤치마크인 hyper-τ-bench를 오픈 소스로 출시했습니다. 이 벤치마크는 \"에이전트 구축\" 작업 시나리오에 초점을 맞추어 장기 자율 에이전트 개발을 위한 평가 기준을 제공합니다.","url":"https://www.aioga.com/ko/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:34.755Z"},"es":{"title":"Sierra open-source hyper-τ-bench: Evalúa agentes capaces de construir agentes","summary":"Actualmente, Sierra ha abierto el hyper-τ-bench, un nuevo benchmark de evaluación de agentes de larga duración diseñado para medir la capacidad de un modelo no solo para actuar como agente, sino también para construir agentes. Este benchmark se centra en el escenario de tarea de \"construir agentes\", proporcionando estándares de evaluación para el desarrollo de agentes autónomos de larga duración.","category":"Industria","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: Evalúa agentes capaces de construir agentes - Aioga Noticias de IA","description":"Actualmente, Sierra ha abierto el hyper-τ-bench, un nuevo benchmark de evaluación de agentes de larga duración diseñado para medir la capacidad de un modelo no solo para actuar com...","url":"https://www.aioga.com/es/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:34.927Z"},"fr":{"title":"Sierra open source hyper-τ-bench : Évalue des agents capables de construire des agents","summary":"Sierra a aujourd’hui rendu open source hyper-τ-bench, un nouveau benchmark d’évaluation d’agents à longue durée conçu pour mesurer la capacité d’un modèle non seulement à agir en tant qu’agent mais aussi à construire des agents. Ce benchmark se concentre sur le scénario de « construction d’agents », fournissant des standards d’évaluation pour le développement d’agents autonomes à longue durée.","category":"Industrie","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open source hyper-τ-bench : Évalue des agents capables de construire des agents - Aioga Actualités IA","description":"Sierra a aujourd’hui rendu open source hyper-τ-bench, un nouveau benchmark d’évaluation d’agents à longue durée conçu pour mesurer la capacité d’un modèle non seulement à agir en t...","url":"https://www.aioga.com/fr/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:43.205Z"},"de":{"title":"Sierra Open-Source-Hyper-τ-Bench: Bewertet Agenten, die in der Lage sind, Agenten zu bauen","summary":"Sierra hat heute hyper-τ-bench als Open Source veröffentlicht, einen neuen Langzeit-Benchmark zur Agentenbewertung, der entwickelt wurde, um die Fähigkeit eines Modells zu messen, nicht nur als Agent zu agieren, sondern auch Agenten zu bauen. Dieser Benchmark konzentriert sich auf das Aufgabenszenario der \"Building Agents\" und bietet Bewertungsstandards für die langfristige autonome Agentenentwicklung.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra Open-Source-Hyper-τ-Bench: Bewertet Agenten, die in der Lage sind, Agenten zu bauen - Aioga KI-News","description":"Sierra hat heute hyper-τ-bench als Open Source veröffentlicht, einen neuen Langzeit-Benchmark zur Agentenbewertung, der entwickelt wurde, um die Fähigkeit eines Modells zu messen,...","url":"https://www.aioga.com/de/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:43.170Z"},"pt-BR":{"title":"Sierra open-source hyper-τ-bench: Avalia agentes capazes de construir agentes","summary":"Hoje, a Sierra abriu o hyper-τ-bench, um novo benchmark de avaliação de agentes de longa duração projetado para medir a capacidade de um modelo não apenas de atuar como agente, mas também de construir agentes. Esse benchmark foca no cenário de tarefa de \"construir agentes\", fornecendo padrões de avaliação para o desenvolvimento de agentes autônomos de longa duração.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: Avalia agentes capazes de construir agentes - Aioga Notícias de IA","description":"Hoje, a Sierra abriu o hyper-τ-bench, um novo benchmark de avaliação de agentes de longa duração projetado para medir a capacidade de um modelo não apenas de atuar como agente, mas...","url":"https://www.aioga.com/pt-BR/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:52.019Z"},"ru":{"title":"Sierra open-source hyper-τ-bench: оценивает агенты, способные создавать агенты","summary":"Сегодня Sierra выпустила с открытым исходным кодом hyper-τ-bench — новый долгосрочный бенчмарк оценки агентов, предназначенный для измерения способности модели не только выступать в роли агента, но и создавать агентов. Этот бенчмарк сосредоточен на сценарии задач «построения агентов», предоставляя стандарты оценки для долгосрочной автономной разработки агентов.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: оценивает агенты, способные создавать агенты - Aioga Новости ИИ","description":"Сегодня Sierra выпустила с открытым исходным кодом hyper-τ-bench — новый долгосрочный бенчмарк оценки агентов, предназначенный для измерения способности модели не только выступать...","url":"https://www.aioga.com/ru/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:21:52.224Z"},"ar":{"title":"منصة سييرا مفتوحة المصدر فائقة τ-bench: تقيم الوكلاء القادرين على بناء الوكلاء","summary":"أما سييرا اليوم فقد أطلقت Hyper-τ-bench مفتوحة المصدر، وهي معيار تقييم وكيل طويل الأمد جديد مصمم لقياس قدرة النموذج ليس فقط على العمل كوكيل بل أيضا على بناء وكلاء. يركز هذا المعيار على سيناريو مهمة \"بناء الوكلاء\"، حيث يوفر معايير تقييم لتطوير الوكلاء المستقلين طويل الأمد.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"منصة سييرا مفتوحة المصدر فائقة τ-bench: تقيم الوكلاء القادرين على بناء الوكلاء - Aioga أخبار الذكاء الاصطناعي","description":"أما سييرا اليوم فقد أطلقت Hyper-τ-bench مفتوحة المصدر، وهي معيار تقييم وكيل طويل الأمد جديد مصمم لقياس قدرة النموذج ليس فقط على العمل كوكيل بل أيضا على بناء وكلاء. يركز هذا المعيار...","url":"https://www.aioga.com/ar/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:00.002Z"},"hi":{"title":"सिएरा ओपन-सोर्स हाइपर-τ-बेंच: एजेंटों के निर्माण में सक्षम एजेंटों का मूल्यांकन करता है","summary":"सिएरा आज ओपन-सोर्स हाइपर-τ-बेंच, एक नया लंबी अवधि का एजेंट मूल्यांकन बेंचमार्क है जिसे न केवल एक एजेंट के रूप में कार्य करने के लिए बल्कि एजेंटों का निर्माण करने के लिए भी मॉडल की क्षमता को मापने के लिए डिज़ाइन किया गया है। यह बेंचमार्क \"बिल्डिंग एजेंटों\" के कार्य परिदृश्य पर केंद्रित है, जो लंबी अवधि के स्वायत्त एजेंट विकास के लिए मूल्यांकन मानक प्रदान करता है।","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"सिएरा ओपन-सोर्स हाइपर-τ-बेंच: एजेंटों के निर्माण में सक्षम एजेंटों का मूल्यांकन करता है - Aioga AI समाचार","description":"सिएरा आज ओपन-सोर्स हाइपर-τ-बेंच, एक नया लंबी अवधि का एजेंट मूल्यांकन बेंचमार्क है जिसे न केवल एक एजेंट के रूप में कार्य करने के लिए बल्कि एजेंटों का निर्माण करने के लिए भी मॉडल की...","url":"https://www.aioga.com/hi/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:00.024Z"},"it":{"title":"Sierra open-source hyper-τ-bench: Valuta agenti in grado di costruire agenti","summary":"Oggi Sierra ha open-sourceato hyper-τ-bench, un nuovo benchmark di valutazione degli agenti a lunga durata progettato per misurare la capacità di un modello non solo di agire come agente, ma anche di costruire agenti. Questo benchmark si concentra sullo scenario operativo di \"costruire agenti\", fornendo standard di valutazione per lo sviluppo autonomo di agenti a lunga durata.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: Valuta agenti in grado di costruire agenti - Aioga Notizie IA","description":"Oggi Sierra ha open-sourceato hyper-τ-bench, un nuovo benchmark di valutazione degli agenti a lunga durata progettato per misurare la capacità di un modello non solo di agire come...","url":"https://www.aioga.com/it/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:08.874Z"},"nl":{"title":"Sierra open-source hyper-τ-bench: Evalueert agenten die agenten kunnen bouwen","summary":"Sierra heeft vandaag hyper-τ-bench open-source gemaakt, een nieuwe langdurige agent evaluatiebenchmark die is ontworpen om het vermogen van een model te meten, niet alleen als agent te fungeren, maar ook om agenten te bouwen. Deze benchmark richt zich op het taakscenario van \"bouwagenten\" en biedt evaluatiestandaarden voor langdurige autonome agentontwikkeling.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: Evalueert agenten die agenten kunnen bouwen - Aioga AI-nieuws","description":"Sierra heeft vandaag hyper-τ-bench open-source gemaakt, een nieuwe langdurige agent evaluatiebenchmark die is ontworpen om het vermogen van een model te meten, niet alleen als agen...","url":"https://www.aioga.com/nl/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:08.949Z"},"tr":{"title":"Sierra açık kaynaklı hyper-τ-bench: Ajanlar oluşturabilen ajanları değerlendirir","summary":"Sierra bugün açık kaynaklı hyper-τ-bench adlı yeni bir uzun vadeli ajan değerlendirme kıyaslası sunuyor; bu kıyas, bir modelin sadece ajan olarak hareket etme değil, aynı zamanda ajanlar oluşturma yeteneğini ölçmek için tasarlandı. Bu kıyaslama, \"ajanlar oluşturma\" görev senaryosuna odaklanır ve uzun vadeli otonom ajan geliştirme için değerlendirme standartları sağlar.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra açık kaynaklı hyper-τ-bench: Ajanlar oluşturabilen ajanları değerlendirir - Aioga AI Haberleri","description":"Sierra bugün açık kaynaklı hyper-τ-bench adlı yeni bir uzun vadeli ajan değerlendirme kıyaslası sunuyor; bu kıyas, bir modelin sadece ajan olarak hareket etme değil, aynı zamanda a...","url":"https://www.aioga.com/tr/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:17.176Z"},"vi":{"title":"Sierra mã nguồn mở hyper-τ-bench: Đánh giá các tác nhân có khả năng xây dựng tác nhân","summary":"Sierra hiện nay cung cấp mã nguồn mở hyper-τ-bench, một chuẩn đánh giá tác nhân mới có thời gian dài được thiết kế để đo lường khả năng của mô hình không chỉ hoạt động như một tác nhân mà còn trong việc xây dựng các tác nhân. Tiêu chuẩn này tập trung vào kịch bản nhiệm vụ của \"xây dựng tác nhân\", cung cấp các tiêu chuẩn đánh giá cho phát triển tác nhân tự chủ trong thời gian dài.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra mã nguồn mở hyper-τ-bench: Đánh giá các tác nhân có khả năng xây dựng tác nhân - Tin tức AI Aioga","description":"Sierra hiện nay cung cấp mã nguồn mở hyper-τ-bench, một chuẩn đánh giá tác nhân mới có thời gian dài được thiết kế để đo lường khả năng của mô hình không chỉ hoạt động như một tác...","url":"https://www.aioga.com/vi/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:17.839Z"},"id":{"title":"Sierra open-source hyper-τ-bench: Mengevaluasi agen yang mampu membangun agen","summary":"Sierra saat ini membuka sumber hyper-τ-bench, sebuah tolok ukur evaluasi agen jangka panjang baru yang dirancang untuk mengukur kemampuan model tidak hanya bertindak sebagai agen tetapi juga membangun agen. Tolok ukur ini berfokus pada skenario tugas \"membangun agen,\" menyediakan standar evaluasi untuk pengembangan agen otonom jangka panjang.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: Mengevaluasi agen yang mampu membangun agen - Berita AI Aioga","description":"Sierra saat ini membuka sumber hyper-τ-bench, sebuah tolok ukur evaluasi agen jangka panjang baru yang dirancang untuk mengukur kemampuan model tidak hanya bertindak sebagai agen t...","url":"https://www.aioga.com/id/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:26.373Z"},"th":{"title":"Sierra open-source hyper-τ-bench: ประเมินเอเจนต์ที่สามารถสร้างเอเจนต์ได้","summary":"ปัจจุบัน Sierra ได้นําเสนอ hyper-τ-bench แบบโอเพ่นซอร์ส ซึ่งเป็นมาตรฐานการประเมินเอเจนต์ระยะยาวรูปแบบใหม่ที่ออกแบบมาเพื่อวัดความสามารถของโมเดลไม่เพียงแต่ทําหน้าที่เป็นเอเจนต์เท่านั้น แต่ยังสามารถสร้างเอเจนต์ได้ด้วยมาตรฐานนี้เน้นไปที่สถานการณ์งานของ \"การสร้างเอเจนต์\" โดยให้มาตรฐานการประเมินสําหรับการพัฒนาเอเจนต์อิสระระยะยาว","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: ประเมินเอเจนต์ที่สามารถสร้างเอเจนต์ได้ - ข่าว AI Aioga","description":"ปัจจุบัน Sierra ได้นําเสนอ hyper-τ-bench แบบโอเพ่นซอร์ส ซึ่งเป็นมาตรฐานการประเมินเอเจนต์ระยะยาวรูปแบบใหม่ที่ออกแบบมาเพื่อวัดความสามารถของโมเดลไม่เพียงแต่ทําหน้าที่เป็นเอเจนต์เท่านั...","url":"https://www.aioga.com/th/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:26.650Z"},"pl":{"title":"Sierra open-source hyper-τ-bench: Ocenia agentów zdolnych do budowania agentów","summary":"Sierra udostępniła dziś otwarte oprogramowanie hyper-τ-bench, nowy benchmark oceny agentów o długim okresie, zaprojektowany do mierzenia zdolności modelu nie tylko do działania jako agent, ale także do budowania agentów. Ten benchmark koncentruje się na scenariuszu zadaniowym \"budowania agentów\", dostarczając standardy oceny dla długoterminowego rozwoju autonomicznych agentów.","category":"行业动态","source":"Sierra：Blog（RSS）","aggregationSource":"Sierra：Blog（RSS）","pageTitle":"Sierra open-source hyper-τ-bench: Ocenia agentów zdolnych do budowania agentów - Aioga Wiadomości AI","description":"Sierra udostępniła dziś otwarte oprogramowanie hyper-τ-bench, nowy benchmark oceny agentów o długim okresie, zaprojektowany do mierzenia zdolności modelu nie tylko do działania jak...","url":"https://www.aioga.com/pl/news/cmtt9xrai020orofps61otpfi/","contentTranslated":true,"sourceHash":"89350e81efbc2558","translatedAt":"2026-09-08T23:22:34.756Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}