--- 格式版本: 2 标题: "Research acceleration: The view inside OpenAI" 原文链接: "https://openai.com/index/research-acceleration-view-inside-openai" 发布日期: "2026-09-06" 发布时间校准状态: "found" 发布时间需复核: "否" 发布时间来源: "llm:local:strict_original_body" 发布时间证据: "September 6, 2026" 发布时间校准原因: "正文中明确标注的发布日期为September 6, 2026，且位于正文开头附近，符合发布时间特征。" 发布时间校准置信度: "1" 发布时间候选数量: 13 发布时间严格候选数量: 4 发布时间原页读取状态: "source template page reused from URL open" 发布时间未找到原因: "" 发布时间校准时间: "2026-09-07T22:14:26+08:00" 发布时间仲裁状态: "confirmed" 发布时间仲裁尝试次数: 1 发布时间仲裁耗时毫秒: 5604 发现时间: "2026-09-07T22:06:43+08:00" 入库时间: "2026-09-07T14:14:42.099Z" 来源平台: "固定入口" 搜索渠道: "fixed_url" 搜索词: "https://openai.com/news/security/" 匹配关键词: - "AI" - "deployment" - "performance" - "GPU" 相关厂家: - "OpenAI" 相关专家: [] 内容类型: "网页" 抓取工具: "Jina Reader" 清洗工具: "Jina Reader Markdown + Defuddle/Readability 正文提取" 原始附件: [] AI优质: "否" AI打分: 35 AI分档: "非优质" AI质检状态: "不通过" AI打分理由: "正文主线为OpenAI内部coding agents对研究流程的加速、使用量统计及安全限制对RL训练compute分配的影响，未直接讨论超节点、AI Rack、机架级系统或关键部件。来源为OpenAI官方一手，内容完整且有新内部数据，但属于研究流程与模型研究效率范畴，无rack-scale硬件、互连、供电、液冷、RAS等基础设施细节，也不构成量产/客户/部署信号。历史对照无关键新增，不命中任何高价值准入通道，并命中应用/研究流程类非机架级基础设施否决。主题相关性0分，技术细节很低，商业部署信号0分，总分35。" AI质检模型: "zj-deepseek-v4-flash" AI质检时间: "2026-09-07T22:17:11+08:00" AI主题相关性: 0 AI来源权威性: 15 AI新颖性: 10 AI技术细节: 1 AI商业部署信号: 0 AI完整性: 9 AI评分提示词版本: "v17-精简生产版" AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d" AI评分知识库版本: "knowledge_base_v1-20260819+runtime.97" AI评分知识库SHA256: "093c22b302b0d377ed12eab9c169d422553a099393d75a2ea63f76a9eb1d8782" AI评分知识库检索词: "[\"OpenAI\",\"https://openai.com/news/security/\",\"RAS\",\"GPU\",\"Intel\",\"URL\",\"RL\",\"API\",\"AGI\",\"x.com/sama/status/1983584366547829073\",\"RSI\",\"NET\"]" AI评分知识库命中: "[{\"id\":\"runtime-51cfe3db04ed0bd3efb5e0e3\",\"title\":\"The full stack behind abundant intelligence\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-25\",\"matchedTerms\":[\"OpenAI\",\"RAS\",\"Intel\",\"URL\",\"RL\",\"API\",\"NET\"],\"rank\":-14.835205664863164},{\"id\":\"july-correct-0081\",\"title\":\"AAI 2026: 6th Gen AMD EPYC Server CPUs Power the Agentic Data Center\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"RAS\",\"GPU\",\"Intel\",\"URL\",\"RL\",\"API\",\"AGI\"],\"rank\":-14.127672724572424},{\"id\":\"runtime-b9406ae170bd91336ab3bb25\",\"title\":\"Jalapeño’s first results show industry-leading speed and efficiency in AI inference\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-25\",\"matchedTerms\":[\"OpenAI\",\"RAS\",\"Intel\",\"URL\",\"RL\",\"API\",\"NET\"],\"rank\":-13.29314812101057},{\"id\":\"july-correct-0094\",\"title\":\"NVIDIA Vera CPU: Olympus Cores Built for Maximum Single-Thread Performance in Agentic AI\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"RAS\",\"GPU\",\"URL\",\"RL\",\"AGI\",\"RSI\",\"NET\"],\"rank\":-12.495720729969353},{\"id\":\"runtime-bf4039a0342d37545e9459a2\",\"title\":\"Most Neoclouds Suck At Security\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-30\",\"matchedTerms\":[\"OpenAI\",\"RAS\",\"GPU\",\"Intel\",\"RL\",\"API\",\"AGI\",\"RSI\",\"NET\"],\"rank\":-10.707449432311254}]" 采集批次: "2026年9月7日22点05分18秒" 采集批次ID: "20260907-220517-361" 去重键: "https://openai.com/index/research-acceleration-view-inside-openai" --- Title: Research acceleration: The view inside OpenAI URL Source: https://openai.com/index/research-acceleration-view-inside-openai Markdown Content: [Skip to main content](https://openai.com/index/research-acceleration-view-inside-openai#main) [](https://openai.com/) * [Research](https://openai.com/research/index/) * Products * [Business](https://openai.com/business/) * [Developers](https://openai.com/api/) * [Company](https://openai.com/about/) * [Foundation(opens in a new window)](https://openaifoundation.org/) Log in[Try ChatGPT(opens in a new window)](https://chatgpt.com/) * Research * Products * Business * Developers * Company * [Foundation(opens in a new window)](https://openaifoundation.org/) [Try ChatGPT(opens in a new window)](https://chatgpt.com/)Login OpenAI September 6, 2026 [Research](https://openai.com/news/research/)[Publication](https://openai.com/research/index/publication/)[Safety](https://openai.com/news/safety-alignment/) # Research acceleration: The view inside OpenAI Loading… [Audio 1](https://openai.com/index/research-acceleration-view-inside-openai) Share 1. Coding agents are reshaping daily work for OpenAI researchers * [1. Coding agents are reshaping daily work for OpenAI researchers](https://openai.com/index/research-acceleration-view-inside-openai#coding-agents-are-reshaping-daily-work-for-openai-researchers) * [2. Researchers are writing more code and running more experiments](https://openai.com/index/research-acceleration-view-inside-openai#researchers-are-writing-more-code-and-running-more-experiments) * [3. The work researchers use agents for is changing](https://openai.com/index/research-acceleration-view-inside-openai#the-work-researchers-use-agents-for-is-changing) * [4. Pacing model development](https://openai.com/index/research-acceleration-view-inside-openai#pacing-model-development) * [5. The path ahead](https://openai.com/index/research-acceleration-view-inside-openai#the-path-ahead) * [Appendix: Our methods for this post](https://openai.com/index/research-acceleration-view-inside-openai#appendix-our-methods-for-this-post) * [1. Coding agents are reshaping daily work for OpenAI researchers](https://openai.com/index/research-acceleration-view-inside-openai#coding-agents-are-reshaping-daily-work-for-openai-researchers) * [2. Researchers are writing more code and running more experiments](https://openai.com/index/research-acceleration-view-inside-openai#researchers-are-writing-more-code-and-running-more-experiments) * [3. The work researchers use agents for is changing](https://openai.com/index/research-acceleration-view-inside-openai#the-work-researchers-use-agents-for-is-changing) * [4. Pacing model development](https://openai.com/index/research-acceleration-view-inside-openai#pacing-model-development) * [5. The path ahead](https://openai.com/index/research-acceleration-view-inside-openai#the-path-ahead) * [Appendix: Our methods for this post](https://openai.com/index/research-acceleration-view-inside-openai#appendix-our-methods-for-this-post) For AGI to benefit all of humanity, we believe it must be democratically governed. This can only happen through an informed public debate about the capabilities, risks and safeguards of highly capable AI systems. People everywhere need to understand the likely future trajectory of frontier AI, so they can have a meaningful voice in how it develops. Transparency about specific risks, incidents and safeguards is necessary, but not sufficient. We believe the public also needs to understand how the most capable systems are developing, and how they are driving research progress, inside of frontier labs. We aim to safely build an automated AI researcher that can work under human supervision to further progress on deep learning and alignment, enabling iterative improvements. According to our measurements, we have now reached the goal, [announced⁠(opens in a new window)](https://x.com/sama/status/1983584366547829073?lang=en) last fall, of having an automated research intern by September of this year. By “research intern,” we mean a system that can carry out well-defined research tasks under human direction, including tasks that would take a skilled researcher a few days. We are making strong progress toward creating an automated AI researcher by March of 2028. Over the course of this year, OpenAI researchers’ daily work has changed substantially. Researchers are using coding agents throughout the day (often in concurrent sessions) and total usage is rapidly increasing, outpacing growth among other OpenAI teams. Researchers are contributing code faster and running more experiments. The ways researchers use agents are changing, too: agents are handling increasingly complex tasks, and succeeding at them more often. AI research is a complex process with many potential bottlenecks, so the overall pace of progress likely won’t keep pace with these specific metrics. But on the whole, these findings are consistent with the broader impression many of us have internally that agentic tools are meaningfully accelerating research progress. People still set our research priorities, judge which ideas and results to pursue, and decide whether to scale, pause, or deploy systems. If it is done responsibly, we believe automated AI research will yield models that directly enhance human welfare and advance OpenAI’s mission. It can bring down the cost of advanced intelligence so that people worldwide can benefit. We are pursuing this work in part because automated research could help us solve alignment and build defenses against increasingly capable AI. An automated AI researcher can also be an automated safety or alignment researcher. More capable, aligned systems could help secure critical infrastructure, defend against dangerous AI agents, and develop new protective measures. These are reasons to develop useful automated research capabilities, but they do not mean that rapid RSI is necessarily an outcome we should pursue. Whether and how to proceed must depend on our ability to preserve human control and on informed democratic choices about the benefits and risks. We do not yet know how to safely get all the way to aligned, full RSI. We are working to scale alignment and safety measures alongside capabilities. But we cannot assume that progress in alignment and safety will keep pace, and more capable systems can become harder to monitor. Careful alignment and safety work is at the center of this effort, and it starts with measuring and mitigating the safety problems we see today in agentic coding systems. Whenever we find that proceeding would pose an unacceptable safety risk, we will respond appropriately including by slowing or stopping our development or deployment of systems we find ourselves unable to sufficiently safeguard. After the recent Hugging Face incident, we [put this commitment into action⁠](https://openai.com/index/pacing-model-development-cyber-capabilities/), pausing reinforcement learning (RL) training on our latest models intended for deployment while we further hardened and red-teamed our research environments and expanded coverage of our monitoring systems. This did not halt all research: some workloads resumed under stronger controls, while others remained paused. We have raised our safety and alignment standards and moved safety work deeper into the model lifecycle, requiring stronger evidence of aligned behavior throughout all of training. Today we are providing a detailed snapshot of how agentic systems have contributed to our progress toward RSI in recent months. Agentic systems are new and rapidly changing, and our measurement efforts are still preliminary. By sharing these early results and the methods behind them, we aim to inform the public, encourage a norm of public disclosure, and help the field move toward shared standards of measurement. Ultimately, as we wrote in our [frontier policy blueprint⁠](https://openai.com/index/frontier-safety-blueprint/), we believe that we and other companies should be required to publicly track our progress toward RSI. Even without such a requirement, we plan to continue being transparent about our RSI progress. We will evolve our transparency approach as our measurement techniques and understanding improve, while balancing the need to protect security and proprietary information. ## 1. Coding agents are reshaping daily work for OpenAI researchers At the start of this year, the median researcher ranked by agent usage at OpenAI was using coding agents only in modest amounts. By mid-August, the median researcher was integrating agents daily into their work, using more than $600 per day of inference at API prices. The 90th percentile user in our research organization now uses more than $7,000 of tokens per day. View methods View methods Before June 2026, total agent runtime across the research organization was still below that of total human labor. That has since changed. In terms of a standard 8 hour workday, as of mid-August, in total, the research organization uses 3.1 agent-workdays of effort for every workday of human labor. View methods Another way of looking at this is to understand how many researchers use highly concurrent workflows (e.g., running 4 or more agents simultaneously). As shown below, this number is increasing. These figures include the daily peaks of both agents started directly by the user and subagents created downstream from those the user launched directly. View methods ## 2. Researchers are writing more code and running more experiments Much of AI research can be seen as a labor-intensive process with the goal of integrating a new improvement to model intelligence or performance into one of our core models. The process depends on many steps, and capabilities advance when all the steps go right together: Researchers have to design new improvements, write evaluations to judge model performance, write infrastructure to test these improvements at scale, catch bugs as well as unsafe or misaligned behavior during training, and integrate winning ideas into a core training run. A failure at any part of the research process can constrain the entire loop. Writing code and running experiments are two major activities that researchers do as part of their work, and we see evidence that these processes are accelerating. View methods These data points are relatively easy to measure, but can be hard to interpret. As automation progresses, the tasks which are _least_ automatable will take on a larger share of researcher effort and will become the important bottlenecks to future progress. Compute is another gating factor for progress, and may become more important over time as other bottlenecks diminish. Through 2026, the number of experiments per active experimenter has increased, with August 2026 being an all-time high since tracking began in Jan 2025. This is correlated with increased Codex adoption, though we note that our available compute has also grown significantly since 2025. View methods ## 3. The work researchers use agents for is changing Both qualitative impressions and internal data indicate that the mix of tasks researchers delegate to coding agents is changing, with delegation of higher level and longer-horizon tasks becoming more common over time. To get a clearer picture of this trend, we analyzed recent usage in the research organization using a [recently published taxonomy⁠(opens in a new window)](https://epoch.ai/gradient-updates/toward-an-onet-for-ai-rnd) of the different kinds of work that are part of the AI R&D lifecycle, developed by Epoch AI. This taxonomy, inspired by the longstanding O*NET system for classifying all kinds of work, is specifically tailored to frontier AI R&D, and breaks the process down into six main phases: 1. Decide: what to work on, what to continue, where to allocate 2. Design: research ideas and engineering specs 3. Build: code and datasets 4. Run: training/eval runs, hardware, serving 5. Analyze: experiments, models, deployment, external work 6. Communicate: findings, feedback, status, decisions Below, we classify coding agent tokens under this taxonomy. View methods View methods We see that all categories of research activities have increased between January and August 2026. In January, the dominant category was research and infrastructure code. This category has expanded, but we also see notable increases in additional categories, especially technical help and monitoring runs. High-level planning still remains a minimal fraction of agent output tokens. Anecdotally, colleagues report that coding agents excel at troubleshooting internal research infrastructure, which addresses one meaningful bottleneck to research progress. Multiple teams which previously held office hours to help researchers troubleshoot their experiments have noted declining attendance in 2026, and one has stopped holding sessions entirely, to focus on making other system improvements instead. Here, we plot the number of top-level posts per day to one of the main internal channels where researchers seek technical support from other teams. To our knowledge, the channel’s decrease in activity has not been offset by queries shifting to another technical support channel run by humans. The decline in traffic aligns with this broader shift. View methods We can also study whether coding agents are succeeding at the tasks researchers request. Using an agentic classifier, we find that from January to July, success rates generally increased across several difficulty buckets (proxied as the estimated time a human would take to complete the task) on tasks we can find a ground truth outcome for. However, agents still require significant human steering to be successful, especially as task complexity rises. In the last 6 months, over half of successful 4-8 hour tasks involved 1 or more interventions. Success rates on researcher tasks have increased over time. Graph excludes classifications where the outcome was uncertain and points with <50 sessions or <50 unique users. View methods Task success and intervention rate from Jan to July, broken out by time horizon. Excludes classifications where the outcome was uncertain. View methods ## 4. Pacing model development Progress toward more capable systems for safe and beneficial AGI will also depend on the safeguards needed for such work. Our assessment of the needed safeguards may change as we learn more about the risks. [As we have described,⁠](https://openai.com/index/pacing-model-development-cyber-capabilities/) we have recently updated our standards for monitoring, alignment, and security. Here, we show how recent restrictions have affected one aspect of research activity. *The majority of Astra compute shown here between July 20 and August 6 was intended to test the implementation of safety and security improvements. View methods On July 20, following the discovery that agents had compromised our research infrastructure, we temporarily shut down the container service used for training, and then restored it with significant additional restrictions. This led to a sharp decline in RL training compute while teams reconfigured their workflows to operate within the hardened research environment. The plot above includes the two week pause in reinforcement learning on our latest models intended for deployment. Astra-class RL experiments between July 20 and August 6 include a majority of runs (by GPU allocation) intended to test the implementation of safety and security improvements. On August 7, preliminary evidence that Astra [may have critical cyber capabilities⁠](https://openai.com/index/responding-next-frontier-critical-cyber-capabilities/) under our [Preparedness Framework⁠(opens in a new window)](https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf) led to additional model-specific security restrictions which required the Astra model to be run in higher security research environments. In the following week, Astra-class GPU allocation fell a further 59.2 percent, but allocation to other model classes rose 17.2 percent. That increase offset about 85 percent of the Astra-class decline, leaving total allocation in the analyzed RL workloads largely unchanged. This pattern is consistent with substitution of some training and experimentation to non-Astra models while work involving Astra was restricted, and comports with anecdotal reports of researchers finding other uses for compute that could no longer be leveraged for workloads covered by the new constraints. This data provides a useful signal for ongoing conversations about training and safety: When new controls are introduced, compute remains valuable and flexible, and will naturally be channeled into alternative uses within the research enterprise. Longer term, discussions about the pace of AI progress should also extend to the question of how compute that is subject to new or proposed controls can best be used. ## 5. The path ahead Making and understanding progress toward aligned RSI is important for our mission. We will continue to refine our methods, report on our evolving understanding, and work toward an informed public debate and meaningful democratic governance of frontier systems. ## Appendix: Our methods for this post Agent-powered AI research is still new, and we are still learning how to measure it. Some indicators, such as the amount of code our research teams generate, are relatively easy to gather, but hard to interpret because their relationship to research progress is uncertain. Metrics that focus more directly on research progress—such as how often agents succeed at the tasks researchers give them—could be more useful, but are complex to develop and validate. Furthering the difficulty, the tools and systems researchers rely on are evolving rapidly. Deepening our understanding of research acceleration is a significant focus area across OpenAI. Across these analyses, unless otherwise noted: * “Researcher” is a broad term for any member of our research organization, including some who build research infrastructure, manage research projects, or otherwise support the enterprise. * Metrics of coding agent use cover most, but not all, usage given rapid evolution in the tools and systems researchers rely on. * [2026](https://openai.com/news/?tags=2026) * [Economic Research](https://openai.com/news/?tags=economic-research) ## Author OpenAI ## Keep reading [View all](https://openai.com/news/) ![Image 1: An alien mind > Listing card](https://images.ctfassets.net/kftzwdyauwt9/7ut3G8rKt5ia4P3yRqi2qN/6ffb429f70548a881eb46a4e7e61498e/Option_120___1080_1080.png?w=3840&q=90&fm=webp) [An Alien Mind Safety Sep 6, 2026](https://openai.com/index/an-alien-mind/) [GPT-6 Astra: A new generation of intelligence Research Sep 3, 2026](https://openai.com/index/gpt-6-astra/) ![Image 2: Safety overview: GPT-6 Astra](https://images.ctfassets.net/kftzwdyauwt9/4vjHRXipk1bBYL1d5Jz11j/d2dbb46d66b1035b7f5f320e818e1608/gpt-6-astra-safety-overview-cover.png?w=3840&q=90&fm=webp) [Safety overview: GPT-6 Astra Safety Sep 3, 2026](https://openai.com/index/safety-overview-gpt-6-astra/) Research * [Research Index](https://openai.com/research/index/) * [Research Overview](https://openai.com/research/) * [Economic Research](https://openai.com/signals/) Latest Advancements * [GPT-6](https://openai.com/index/gpt-6-astra/) * [GPT-5.6](https://openai.com/index/gpt-5-6/) * [GPT-5.5](https://openai.com/index/introducing-gpt-5-5/) * [GPT-5.4](https://openai.com/index/introducing-gpt-5-4/) Safety * [Safety Approach](https://openai.com/safety/) * [Deployment Safety(opens in a new window)](https://deploymentsafety.openai.com/) * [Security & Privacy](https://openai.com/security-and-privacy/) * [Trust & Transparency](https://openai.com/trust-and-transparency/) Products * [ChatGPT(opens in a new window)](https://chatgpt.com/) * [ChatGPT Business(opens in a new window)](https://chatgpt.com/business/) * [ChatGPT Enterprise(opens in a new window)](https://chatgpt.com/business/enterprise/) * [ChatGPT for Education(opens in a new window)](https://chatgpt.com/business/education/) * [Codex](https://openai.com/codex/) * [Release Notes](https://openai.com/products/release-notes/) API Platform * [Overview](https://openai.com/api/) * [API Log In(opens in a new window)](https://platform.openai.com/login) * [Docs(opens in a new window)](https://developers.openai.com/api/docs) Business * [Overview](https://openai.com/business/) * [Solutions](https://openai.com/solutions/) * [Resources](https://openai.com/business/learn/) * [Customer Stories](https://openai.com/business/customer-stories/) * [Partner Network](https://openai.com/business/partners/) * [Contact Sales](https://openai.com/contact-sales/) Developers * [Apps SDK(opens in a new window)](https://developers.openai.com/apps-sdk) * [Open Models](https://openai.com/open-models/) * [Docs(opens in a new window)](https://developers.openai.com/) * [Resources(opens in a new window)](https://developers.openai.com/learn) * [Developer Forum(opens in a new window)](https://community.openai.com/) Company * [About Us](https://openai.com/about/) * [Our Charter](https://openai.com/charter/) * [Careers](https://openai.com/careers/) * [News](https://openai.com/news/) Support * [Help Center(opens in a new window)](https://help.openai.com/) More * [Stories](https://openai.com/stories/) * [Academy](https://openai.com/academy/) * [Supply Co.](https://openai.com/supply/) * [Livestreams](https://openai.com/live/) * [Podcast](https://openai.com/podcast/) * [RSS](https://openai.com/news/rss.xml) Terms & Policies * [Terms of Use](https://openai.com/policies/terms-of-use/) * [Privacy Policy](https://openai.com/policies/privacy-policy/) * [Other Policies](https://openai.com/policies/) [(opens in a new window)](https://x.com/OpenAI)[(opens in a new window)](https://www.youtube.com/OpenAI)[(opens in a new window)](https://www.linkedin.com/company/openai)[(opens in a new window)](https://github.com/openai)[(opens in a new window)](https://www.instagram.com/openai/)[(opens in a new window)](https://www.tiktok.com/@openai)[(opens in a new window)](https://discord.gg/openai) OpenAI © 2015–2026 Your privacy choices English United States