[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-claude-haiku-5-5-launch":3,"topics-all":39,"news-related-a5c82cbd-def9-4d33-99b3-cba1d7ab4abe":58},{"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":37,"view_count":38},"a5c82cbd-def9-4d33-99b3-cba1d7ab4abe","Haiku 5.5 发布:输入一毛钱,跑分最高翻4.6倍","Anthropic 发布小模型 Claude Haiku 5.5:输入每百万 token 0.1 美元,10 万 token 内较上代降九成;OSWorld 离线子集 15.7% 升至 72.4%,Terminal-Bench 4.0 从 0 到 39.2%,首次支持 effort 分档,上下文规格标至 1M。","Anthropic 用一个月凑齐了 5.5 家族:9 月 22 日 Opus 5.5,9 月 28 日 Sonnet 5.5,10 月 7 日轮到最小的 Haiku。官方称之为「我们发布过的最便宜、最快、最强小模型」——输入价格压到每百万 token 0.1 美元,上代 Haiku 4.5 是 1 美元。\n\n## 价格:九成降幅里藏着一道 10 万 token 分水岭\n\n定价是两档:prompt 在 10 万 token 以内,输入 0.1 美元、输出 0.5 美元;超过 10 万,输入 0.5 美元、输出 2.5 美元。官方脚注很直白:10 万以内降 90%,以外降 50%——而上一代约 90% 的请求都落在 10 万以内,平均「约便宜 75%」。\n\n两个细节值得抠。一是缓存读取低至每百万 0.01 美元,是 agent 负载的杠杆;二是换了新 tokenizer,同一任务消耗的 token 略有增加——第三方通讯刊 roo 发布当天按官方牌价重算,同样文本 10 万线以下实际降约 87%、线以上约 35%,「降九成」要把脚注一起读完才成立。\n\n## 跑分:小模型的跳跃比大模型猛\n\n官方跑分表里,Haiku 5.5 对上代是跨越式的:OSWorld 2.1 离线子集从 15.7% 到 72.4%(GPT-6 Luna 为 48.9%),Humanity's Last Exam 从 10.2% 到 45.9%,Terminal-Bench 4.0 从 0% 到 39.2%,FrontierCode 1.1 拿下 46.4%、压过 GPT-6 Luna 的 42.4%,知识工作 GDPval-AA v2.1 从 735 涨到 1620。\n\n天花板也标得清楚:同表 Sonnet 5.5 对照列是 1840、Terminal-Bench 70.6%。Anthropic 自己承认复杂 agent 编码仍是 Sonnet\u002FOpus 的地盘,Haiku 的主场是压缩、摘要、分类、子代理这类「以前跑不起」的窄任务。\n\n## 用法高度一致:当 subagent 使\n\n早期客户证言清一色把 Haiku 5.5 当子代理:Asana 称任务延迟降三成以上、单轮推理最快快 2.5 倍;HubSpot 在 CRM 评测拿到 92.8% 历史最佳;AlphaSense 文档问答每周 800 万次调用,质量分 0.76 升到 0.84;Cognition 把它塞进 Devin Fusion 当 sidekick,FrontierCode 守在 66.2。它也是首个支持 effort 分档的 Haiku,从 low 到 max 五档;第三方规格页与 Artificial Analysis 均标注 1M 上下文、128K 输出。\n\n## 同场发布的另外三件事\n\nSonnet 5.5 缓存读取从 0.2 降到 0.1 美元,官方估算多数 agent 负载再省约 20%;Max\u002FTeam 订阅新增每月 API 额度(5x 档 100 美元、20x 档 200 美元、Team 池化 500 美元);Python\u002FTypeScript SDK 开始 beta 支持电脑操作与浏览器使用——正是为 Haiku 的速度加价格组合准备的实时交互场景。\n\n所以呢:单看「一毛钱输入」是降价新闻,连起来看是 agent 经济学——当子代理的边际成本跌一个数量级,「一个大模型带一群小模型干活」从架构审美变成财务常识。想迁移的团队,先把 10 万 token 分水岭和 tokenizer 换代的账算清楚。\n\n参考:[Anthropic 官方公告](https:\u002F\u002Fwww.anthropic.com\u002Fclaude-haiku-5-5) · roo.beehiiv.com","https:\u002F\u002Fwww.anthropic.com\u002Fclaude-haiku-5-5","1001cead-0639-4c04-ab47-f19d863be5f2",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"23544f6a-eea1-4f05-aa8d-749ca862d5d2","anthropic",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"dca4d0ab-7994-43a7-839e-7756fc77344a","claude",{"id":19,"name":20,"slug":20,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":22,"name":23,"slug":23,"description":14,"color":14},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"664fbf49-4441-4580-89f1-1580c095710a","en","Claude Haiku 5.5: $0.10 Input, Benchmarks Up to 4.6x","Anthropic ships Claude Haiku 5.5: $0.10\u002FM input (up to 90% cheaper), 1M context, OSWorld 15.7%→72.4%, first Haiku with adjustable effort levels.","Anthropic has filled out the 5.5 family in under a month: Opus 5.5 on September 22, Sonnet 5.5 on September 28, and now the smallest sibling. On October 7, Claude Haiku 5.5 went live — billed by Anthropic as \"the cheapest, fastest, and most capable small model we've ever released,\" with input pricing pushed down to $0.10 per million tokens, versus $1.00 for Haiku 4.5.\n\n## Pricing: a 100K-token watershed hides inside the 90% cut\n\nThe pricing is two-tier: prompts up to 100,000 tokens cost $0.10 per million input and $0.50 per million output; above that line, $0.50 and $2.50. The official footnote is blunt: 90% cheaper under 100K, 50% cheaper over it — and about 90% of requests to the previous Haiku fell under the line, averaging out to \"around 75% cheaper.\"\n\nTwo details deserve scrutiny. First, caching: the under-100K tier prices cache reads at $0.01 per million and writes at $0.125 — a real lever for repetitive agent workloads. Second, the tokenizer change: Haiku 5.5 adopts the updated tokenizer shared with Sonnet 5.5 and Opus 5.5, consuming slightly more tokens per task. The third-party newsletter roo re-ran Anthropic's published prices on launch day and found the real cut is about 87% below the line and about 35% above it — \"90% cheaper\" is a story that only holds if you read the footnote too.\n\n## Benchmarks: the small model jumps harder than the big ones\n\nOn Anthropic's own table, Haiku 5.5 posts order-of-magnitude gains over its predecessor: the OSWorld 2.1 offline subset (real computer use) jumps from 15.7% to 72.4%, with GPT-6 Luna at 48.9%; Humanity's Last Exam without tools goes from 10.2% to 45.9%; Terminal-Bench 4.0 goes from 0% to 39.2%; on FrontierCode 1.1 it scores 46.4%, ahead of GPT-6 Luna's 42.4%. On knowledge work, GDPval-AA v2.1 rises from 735 to 1620, overtaking GPT-6 Luna's 1437.\n\nThe ceiling is marked clearly too: Sonnet 5.5 sits at 1840 in the same table's reference column (83.9% OSWorld, 70.6% Terminal-Bench). Anthropic itself concedes that complex agentic coding remains Sonnet\u002FOpus territory; Haiku's home ground is narrowly scoped tasks that used to be cost-prohibitive — compaction, summarization, classification, subagent work.\n\n## Customers: cheap enough to run without thinking\n\nEarly customers use it almost uniformly as a subagent. Asana reports over 30% lower task-completion latency and up to 2.5x faster inference per agent turn; HubSpot logged 92.8% — its best score yet — on a CRM eval suite; AlphaSense's document Q&A runs 8 million calls a week and improved from 0.76 to 0.84; Box saw 11 points higher scores at about half the latency; Cognition dropped it into Devin Fusion as the sidekick, holding a top-tier FrontierCode score of 66.2. Haiku 5.5 is also the first Haiku-class model with an adjustable effort setting — the official charts label five levels from low to max — letting you tune cost versus intelligence per request. On specs, third-party pages and Artificial Analysis both list a 1M context window and 128K max output, closing the gap with Haiku 4.5's 200K.\n\n## Three more changes shipped alongside\n\nOne: Sonnet 5.5 cache reads drop from $0.20 to $0.10 per million, which Anthropic estimates makes most agentic workloads about 20% cheaper. Two: Max and Team subscribers get a new monthly API credit — $100 for Max 5x, $200 for Max 20x, up to $500 pooled for Team. Three: the Python and TypeScript SDKs add beta support for computer use and browser use — exactly the real-time, interactive workloads Haiku 5.5's speed-plus-price combination was built for.\n\nSo what? Read alone, \"ten-cent input\" is a price-cut headline. Read together, it's agent economics: when the marginal cost of a subagent drops by an order of magnitude, \"one big model directing a fleet of small ones\" stops being an architecture aesthetic and becomes financial common sense. Teams planning to migrate should first do the math on the 100K watershed and the tokenizer swap.\n\nReferences: [Anthropic announcement](https:\u002F\u002Fwww.anthropic.com\u002Fclaude-haiku-5-5) · [roo pricing analysis](https:\u002F\u002Froo.beehiiv.com\u002Fp\u002Fclaude-haiku-5-5-pricing-benchmarks)","claude-haiku-5-5-launch","2026-10-07T21:09:42Z","2026-10-07T21:09:52.797174Z","2026-10-07T21:09:52.797184Z",true,"agent","https:\u002F\u002Fwww-cdn.anthropic.com\u002Fimages\u002F4zrzovbb\u002Fwebsite\u002F89e9b0fbfe0181f6ca565d913686ce8e49b5e9c2-1200x630.jpg",313,[40,49],{"slug":41,"tag_slug":41,"title_zh":42,"title_en":43,"intro_zh":44,"intro_en":45,"id":46,"is_active":35,"created_at":47,"modified_at":48},"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":50,"tag_slug":50,"title_zh":51,"title_en":52,"intro_zh":53,"intro_en":54,"id":55,"is_active":35,"created_at":56,"modified_at":57},"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":59},[60,65,70,75,80,85],{"id":61,"title":62,"news_slug":63,"published_at":64},"6cd8c9b7-b493-4ad7-bd20-a907b9fb70f4","Claude 5.5 家族首发:Opus 5.5 缓存读取降六成,典型负载省四成","claude-opus-5-5-cost-cut-benchmarks","2026-09-22T21:10:00+00:00",{"id":66,"title":67,"news_slug":68,"published_at":69},"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","claude-opus-5-launch","2026-07-25T02:00:00+00:00",{"id":71,"title":72,"news_slug":73,"published_at":74},"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":76,"title":77,"news_slug":78,"published_at":79},"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":81,"title":82,"news_slug":83,"published_at":84},"a7a8ca05-6134-4b27-8dc0-ae3c0a764630","Claude 自主设计量子电路:53 小时拿到一个数量级提速","claude-quantum-circuit-redesign","2026-10-19T10:00:00+00:00",{"id":86,"title":87,"news_slug":88,"published_at":89},"2c1d70a6-02aa-4b17-9413-a8663216baf6","Anthropic 网络验证改三层:Glasswing 并入","anthropic-glasswing-cyber-verification-tiers","2026-10-15T09:10:00+00:00"]