[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-claude-opus-5-launch":3,"topics-all":36,"news-related-5930aa08-8a81-4c92-9ec0-79c3f4de39e2":55},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"5930aa08-8a81-4c92-9ec0-79c3f4de39e2","Anthropic 推出 Claude Opus 5:性能逼近 Fable 5,API 价格不变继续啃企业市场","7月25日,Anthropic 正式发布旗舰模型 Claude Opus 5。新版本在编程、Agent 任务与复杂推理上整体性能逼近更高端的 Claude Fable 5,但 API 价格仍维持上代 Opus 4.8 水平:每百万输入 Token 5 美元、输出 Token 25 美元,定价直接腰斩于 Fable 5。Opus 5 已向 Claude 平台与 API 用户全面开放,主打企业办公、软件开发等高频场景,并进一步增强安全性与抗滥用能力。这次发布释放的信号很清晰:Anthropic 正在用 Opus 系列做主力旗舰下沉,把 Fable 5 这一档能力压到中价位档,让企业用户能用 Opus 4.8 的价格拿到接近 Fable 5 的智能。同行此前已经走过这条路——Anthropic 7 月初把 Sonnet 5 拉到 Opus 4.8 水位,把主力型号从 Opus 切到 Sonnet;Opus 5 这次是反向再下一城。更值得关注的是 Opus 5 与今年陆续上线的 Opus 4.6、4.7、4.8 形成的高频迭代节奏。41 天一个版本的背后,是 Anthropic 在 RL、Agent harness、token 效率上的工程化能力集中外溢。从 4.5 夺回编程榜首、4.7 的 tokenizer 升级、4.8 的 Dynamic Workflows,到这次 Opus 5 的价格不变能力跃迁,每一代都在强化高性价比与高频迭代两条主线。对开发者与企业的实际影响有三层:第一,Agent 工作流与长程任务的成本门槛进一步降低,过去只能用 Fable 5 跑的关键路径可以平替到 Opus 5;第二,Opus 4.8 时代训练好的模型编排、prompt cache 策略可以无缝迁移;第三,Anthropic 主动释放的价格信号会再次搅动闭源 API 市场——Claude、GPT、Gemini、GLM 在中高端档的价格战会变得更难看。所以这次发布的真正重点不是 Opus 5 又强了多少,而是 Anthropic 正在把高端模型的能力,按月为单位向下摊薄。当 Fable 5 和 Opus 5 的能力差不到 10% 而价格差两倍多,企业选型的天平会明显向 Opus 倾斜——这才是 Anthropic 想看到的。","https:\u002F\u002F36kr.com\u002Fp\u002F3910351337821318?f=rss","1fa87d30-d9f3-4752-b3be-0373933b3aaf",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"23544f6a-eea1-4f05-aa8d-749ca862d5d2","anthropic",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"dca4d0ab-7994-43a7-839e-7756fc77344a","claude",{"id":18,"name":19,"slug":19,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":21,"name":22,"slug":22,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":28},"1f5f8c6e-5fe2-4b04-a9f4-589cdd363086","en","Claude Opus 5 launches: near Fable 5 at the same API price","On July 25, Anthropic officially released its flagship Claude Opus 5. The new version overall performance — in coding, Agent tasks, and complex reasoning — approaches the more premium Claude Fable 5, but API pricing stays at last-gen Opus 4.8 levels: $5 per million input tokens, $25 per million output tokens — directly half the price of Fable 5. Opus 5 is now fully open to Claude platform and API users, targeting high-frequency scenarios like enterprise office work and software development, with further enhancements to safety and anti-abuse. The signal is clear: Anthropic is using the Opus line to bring the flagship down, packing Fable 5-class capability into the mid-price tier so enterprise users can get near-Fable 5 intelligence at Opus 4.8 prices. Peers have already walked this path — Anthropic at the start of July pulled Sonnet 5 up to Opus 4.8's level, shifting the workhorse model from Opus to Sonnet; Opus 5 is now a tier-down in the other direction. Also worth noting is the high-frequency iteration cadence that Opus 5 forms with the earlier-shipped Opus 4.6, 4.7, and 4.8. One version every 41 days — that's Anthropic's engineering capability in RL, Agent harness, and token efficiency spilling out in concentrated form. From 4.5 retaking the coding crown, 4.7's tokenizer upgrade, 4.8's Dynamic Workflows, to Opus 5's price-unchanged capability leap, each generation strengthens the two main lines of high cost-effectiveness and high-frequency iteration. There are three layers of practical impact for developers and enterprises. First, the cost barrier for Agent workflows and long-horizon tasks is further lowered — key paths that previously only Fable 5 could run can be substituted with Opus 5. Second, the model orchestration and prompt cache strategies trained in the Opus 4.8 era migrate seamlessly. Third, the price signal Anthropic is actively sending will stir up the closed-source API market again — the price war at the mid-to-high end among Claude, GPT, Gemini, and GLM will only get uglier. So the real point of this release isn't how much stronger Opus 5 is — it's that Anthropic is amortizing the high-end model's capability downward on a monthly cadence. When Fable 5 and Opus 5 differ by less than 10% in capability but more than 2x in price, the enterprise selection scales will obviously tip toward Opus — and that's exactly what Anthropic wants to see.","claude-opus-5-launch","2026-07-25T02:00:00Z","2026-07-25T02:04:12.782393Z","2026-08-19T02:08:40.142862Z",true,"agent",173,[37,46],{"slug":38,"tag_slug":38,"title_zh":39,"title_en":40,"intro_zh":41,"intro_en":42,"id":43,"is_active":33,"created_at":44,"modified_at":45},"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":47,"tag_slug":47,"title_zh":48,"title_en":49,"intro_zh":50,"intro_en":51,"id":52,"is_active":33,"created_at":53,"modified_at":54},"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":56},[57,62,67,72,77,82],{"id":58,"title":59,"news_slug":60,"published_at":61},"20904504-4450-4f2e-a612-49b191403a86","Claude Opus 5 thinking 默认开启：OSWorld 70.57% 的慢即是快","claude-opus-5-thinking-default","2026-07-25T00:00:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"428c2822-a53a-4075-80ec-2960e03ea062","Anthropic Claude Sonnet 5：中端档拉到 Opus 4.8 水位","claude-sonnet-5-main-model","2026-07-01T02:00:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"390c2437-4e4f-45ec-8270-67c5bfa4fa47","ChatGPT、Claude、Grok、Gemini 罕见同时下线,周四早晨全球 AI 集体失声","chatgpt-claude-grok-gemini-thursday-outage","2026-09-05T06:00:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"63b28ef5-3ffa-4897-88d2-5dcd7fb678b5","Claude Fable 5.1 发布:缓存读取降价 75%,Agent 科研基准翻倍","claude-fable-5-1-mythos-release","2026-09-02T13:20:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"f3d17d45-e1a8-4a1b-9449-6813aff06e49","Anthropic 让 Claude 自己修对齐:10 类失败全部见效,还超过人类研究员","claude-automated-alignment-researchers","2026-08-29T13:05:00+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"39724847-fdc9-4199-ac46-311e7b49d385","Ramp 数据复盘 Fable 5:旗舰上市两月仅占企业 Anthropic 支出 11%,70 倍价差压住前沿模型溢价","ramp-data-fable-5-adoption-plateaus","2026-08-26T08:00:00+00:00"]