[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-gpt-6-1-terra-computer-use":3,"topics-all":38,"news-related-ce0faacb-378d-4478-bd14-99cccc140ea0":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"ce0faacb-378d-4478-bd14-99cccc140ea0","GPT-6.1 Terra 首发:十分之一价格追平Astra","OpenAI 发布 GPT-6.1 Terra，主打视觉推理与电脑操作，API 定价较前代砍半。官方评测称其追平 GPT-6 Astra，每 token 价格仅十分之一；幻觉错误率从 53.6% 降至 16.7%，Terminal-Bench 4.0 翻倍至 50.3%。","OpenAI 于 10 月 26 日发布 GPT-6.1 系列新成员 Terra，定位很明确：快速、高效，把细致的视觉推理和精确的电脑操作（computer use）结合起来，写作也更自然。官方将其描述为对 GPT-5.6 Terra 的一次显著升级，瞄准的是填满工作日的任务——起草提案、解读技术图纸、填写表单、剪辑视频。\n\n## 性价比是这次的主叙事\n\nTerra 的标准 API 定价比 GPT-5.6 Terra 和 GPT-6.1 Sol 都低 50%：每百万输入 token 1 美元、缓存输入 0.05 美元、输出 5 美元。在官方发布的评测里，几组数字值得注意：\n\n- 在 Agents' Last Exam（一组职业级任务的电脑操作基准）上，中等推理强度下 Terra 比 GPT-5.6 Terra 高 15 个百分点，成本约为三分之一；与 GPT-6 Astra 表现相当，而每 token 价格只有后者的十分之一。\n- 最大推理强度下，Terra 的 token 效率超过 Claude Opus 5.5 的三倍，单价为其四分之一，折算单任务成本约低 9 倍；完成任务速度比 GPT-6 Astra 和 GPT-6.1 Sol 快三倍以上。\n- 视觉推理基准 MMMU-Pro 上，Terra 在所有推理档位都超过 GPT-5.6 Terra；桌面操作基准 OSWorld 2.1 上，中等强度就以远低于前代的单任务成本刷新了 GPT-5.6 Terra 的最好成绩。\n\n## 事实性与编码：翻倍，但有边界\n\nOpenAI 称今年重点投入了降低幻觉。在内部错误诱导提示基准上，低推理强度的错误率从 GPT-5.6 Terra 的 53.6% 降到 16.7%；最大强度下降 18 个百分点至 4.3%，接近 GPT-6 Astra 和 GPT-6.1 Sol 的水平。\n\n编码方面，Terminal-Bench 4.0 最大强度下得分 50.3%，是 GPT-5.6 Terra 公开分数 21.5% 的两倍多。但官方也明说：最复杂的编码任务，GPT-6.1 Sol 和 GPT-6 Astra 仍是更好的选择——这句边界说明反而值得注意。在科学任务基准 Terminal-Bench-Science 0.1 上，Terra 超过 GPT-6.1 Sol，距 Claude Opus 5.5 的成绩在 5 分以内，而估算成本不到其五分之一。\n\n安全方面，GPT-6.1 Terra 满足 OpenAI Preparedness Framework 网络安全维度的 Critical 门槛，保留了 GPT-6 Astra 的防护措施，系统卡已同步公开。模型即日起面向 ChatGPT Work 和 Codex 的 Plus、Pro、Enterprise、Edu 用户推出，API 名为 gpt-6.1-terra。\n\n## 所以呢\n\n通读发布文，Terra 的真正卖点不是\"最强\"，而是\"每美元能干多少活\"。当电脑操作能力追平旗舰、每 token 单价却只有十分之一时，读发票录系统、批量填表、盯屏幕跑流程这类高频低容错的工作流，终于有了能算得过账的自动化成本。模型竞争的重心正在从跑分榜转向单位经济学——对开发者来说，这比榜单头衔更实际：你的 Agent 下个月烧多少钱，取决于你会不会换模型。\n\n原始发布文：https:\u002F\u002Fopenai.com\u002Findex\u002Fintroducing-gpt-6-1-terra\u002F\n","https:\u002F\u002Fopenai.com\u002Findex\u002Fintroducing-gpt-6-1-terra\u002F","bd0e0e04-6bcf-4b3e-9a56-62c672308ec9",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"6ad31a14-c0da-42df-81fd-564281f768db","agentic-ai",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"baf131c1-687a-49f4-87f6-4dd87c1c692f","gpt",{"id":19,"name":20,"slug":20,"description":14,"color":14},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":22,"name":23,"slug":23,"description":14,"color":14},"42e59a88-7795-47dc-a334-ef1e72c24347","openai",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"59dc0886-b0cb-48a6-8dbc-528974535fa9","en","OpenAI launches GPT-6.1 Terra: cheap, fast computer use","OpenAI's GPT-6.1 Terra pairs visual reasoning with computer use at half GPT-6.1 Sol's API price, matching GPT-6 Astra at one-tenth per-token cost.","OpenAI released GPT-6.1 Terra on October 26, the newest member of its GPT-6.1 series. The positioning is clear: a fast, efficient model that combines detailed visual reasoning with precise computer use, alongside clearer and more natural writing. OpenAI describes it as a significant step up from GPT-5.6 Terra, aimed at the tasks that fill the workday — drafting proposals, interpreting technical drawings, filling out forms, and editing videos.\n\n## Price-performance is the story\n\nTerra's standard API prices are 50% lower than both GPT-5.6 Terra and GPT-6.1 Sol: $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens. Several numbers from the official evaluations stand out:\n\n- On Agents' Last Exam, a benchmark of professional computer-use tasks, Terra at medium reasoning effort scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost, and performs on par with GPT-6 Astra at only one-tenth its price per token.\n- At maximum reasoning effort, Terra is over three times as token-efficient as Claude Opus 5.5, at one-fourth the price per token — roughly a 9x lower cost per task. It also completes tasks over three times faster than both GPT-6 Astra and GPT-6.1 Sol.\n- On MMMU-Pro, a visual reasoning benchmark, Terra outperforms GPT-5.6 Terra at every reasoning setting; on OSWorld 2.1, which tests desktop-app agents, it surpasses GPT-5.6 Terra's best score at a fraction of the cost per task.\n\n## Factuality and coding: doubled, with honest limits\n\nOpenAI says it has invested heavily this year in reducing hallucinations. On its internal benchmark of error-inducing prompts, the error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra; at maximum effort it dropped 18 percentage points to 4.3%, nearly matching GPT-6 Astra and GPT-6.1 Sol.\n\nOn the coding side, Terra solves 50.3% of tasks on Terminal-Bench 4.0 at maximum reasoning effort — more than double GPT-5.6 Terra's published score of 21.5%. But OpenAI is explicit that for the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice — a boundary statement worth noting. On Terminal-Bench-Science 0.1, Terra surpasses GPT-6.1 Sol and comes within five points of Opus 5.5 at less than one-fifth the estimated cost.\n\nOn safety, Terra meets the Critical threshold in cybersecurity under OpenAI's Preparedness Framework, keeps the protections developed for GPT-6 Astra, and ships with a published system card. It is rolling out now to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and is available through the API as gpt-6.1-terra.\n\n## So what\n\nRead the announcement end to end and Terra's real pitch is not \"strongest model\" but \"most work per dollar\". When computer-use capability matches the flagship at one-tenth the per-token price, high-volume, low-tolerance workflows — reading invoices into accounting systems, batch-filling forms, watching screens to run processes — finally have automation economics that add up. The center of gravity in model competition is shifting from leaderboard scores to unit economics, and for developers that is more practical than any title: how much your agents burn next month depends on whether you know how to switch models.\n\nOriginal announcement: https:\u002F\u002Fopenai.com\u002Findex\u002Fintroducing-gpt-6-1-terra\u002F\n","gpt-6-1-terra-computer-use","2026-10-26T19:08:07Z","2026-10-26T19:08:11.733361Z","2026-10-26T19:08:11.733373Z",true,"agent",486,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"746c4779-5fe0-4d4b-a2cb-e3cba576372a","GPT-6 Cyber上线:948个内核提权背后的模型","openai-gpt-6-cyber-daybreak-red","2026-11-06T21:09:45+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"9dffcd24-e154-4986-baab-be712e6e0442","GPT-5.6 Sol通关Factorio:低思考档打出火箭","gpt-5-6-sol-factorio-rocket-run","2026-11-01T13:20:38+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"589f97e0-2e8f-4b3d-b665-3a05ec7ffd56","GPT-6 全民分发:Intelligent UI 让 ChatGPT 学会拼界面","openai-gpt-6-intelligent-ui-global","2026-10-08T03:00:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"cddf2879-ceda-40df-9686-a0e4918d487b","GPT-6.1 Astra 被砍:模型会撒谎,Sol 顶上","openai-cancels-gpt-6-1-astra","2026-09-29T19:12:22+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"03995aa0-5eb5-4044-9b71-95de7f9081ce","OpenAI 上线 Dots:24×7 全时 Agent","openai-dots-agent-launch","2026-09-29T08:00:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"a574d32a-c8f7-4d46-ac95-b76406f36b3b","OpenAI 把 GPT-6 拆成两档:Sol\u002FLuna 同款训练法,API 价格砍一半","openai-gpt-6-sol-luna-api-pricing","2026-09-23T00:00:00+00:00"]