{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-28T06:20:51.496Z","headline":"你调用的 AI 模型可能不是你付钱的那个--Anthropic 被曝静默切换模型","description":"调用 API 时指定 `model： \"claude-fable-5\"`，返回的响应中 `model` 字段却显示 `\"claude-opus-4-8\"`。Anthropic 在生成开始前对请求进行分类，匹配到敏感类别后静默切换至另一组权重，且不报错、不重试。用户实际获得的模型与付费指定的模型不一致。","url":"https://www.aioga.com/news/cmrxueg0q02ryroxpw82u6uld/","mainEntityOfPage":"https://www.aioga.com/news/cmrxueg0q02ryroxpw82u6uld/","datePublished":"2026-07-23T18:07:55.000Z","dateModified":"2026-07-23T18:07:55.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://www.marktechpost.com/2026/07/23/you-didnt-get-the-ai-model-you-paid-for","https://aihot.virxact.com/items/cmrxueg0q02ryroxpw82u6uld"],"canonicalUrl":"https://www.aioga.com/news/cmrxueg0q02ryroxpw82u6uld/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：调用 API 时指定 `model： \"claude-fable-5\"`，返回的响应中 `model` 字段却显示 `\"claude-opus-4-8\"`。 Aioga 将其归入「技巧观点」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/cmrxueg0q02ryroxpw82u6uld/","dateCreated":"2026-07-23T18:07:55.000Z","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":"marktechpost.com source article","url":"https://www.marktechpost.com/2026/07/23/you-didnt-get-the-ai-model-you-paid-for","datePublished":"2026-07-23T18:07:55.000Z","provider":{"@type":"Organization","name":"marktechpost.com","url":"https://www.marktechpost.com/2026/07/23/you-didnt-get-the-ai-model-you-paid-for"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmrxueg0q02ryroxpw82u6uld","datePublished":"2026-07-23T18:07:55.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmrxueg0q02ryroxpw82u6uld"}}],"aggregationSource":"MarkTechPost（RSS）","originalPublisher":{"name":"marktechpost.com","url":"https://www.marktechpost.com/2026/07/23/you-didnt-get-the-ai-model-you-paid-for"},"article":{"id":"cmrxueg0q02ryroxpw82u6uld","slug":"cmrxueg0q02ryroxpw82u6uld","url":"https://www.aioga.com/news/cmrxueg0q02ryroxpw82u6uld/","title":"你调用的 AI 模型可能不是你付钱的那个--Anthropic 被曝静默切换模型","title_en":"You Didn't Get the AI Model You Paid For","summary":"调用 API 时指定 `model： \"claude-fable-5\"`，返回的响应中 `model` 字段却显示 `\"claude-opus-4-8\"`。Anthropic 在生成开始前对请求进行分类，匹配到敏感类别后静默切换至另一组权重，且不报错、不重试。用户实际获得的模型与付费指定的模型不一致。","source":"MarkTechPost（RSS）","sourceUrl":"https://www.marktechpost.com/2026/07/23/you-didnt-get-the-ai-model-you-paid-for","aiHotUrl":"https://aihot.virxact.com/items/cmrxueg0q02ryroxpw82u6uld","publishedAt":"2026-07-23T18:07:55.000Z","category":"技巧观点","score":45,"selected":false,"articleBody":["You call the API. You pass model: “claude-fable-5”. You get back a completion, a token count, and a field that reads “model”: “claude-opus-4-8”.","Nothing errored. Nothing retried. The request was classified before generation began, matched a sensitive category, and was handed to a different set of weights entirely. Anthropic documented this when it brought Fable 5 back on July 1: blocked requests are sent to Opus 4.8 instead, and the user is notified. The switch happens at the API layer, and the response object names the model that actually ran.","That is the well-behaved version. It is also, as far as I can tell, the only version in wide deployment that tells you the truth in-band.","Two weeks later, Cursor shipped Router: a classifier trained on 600,000-plus live requests that reads each query’s context, complexity, and domain and dispatches it to whichever model it judges best with three early-access accounts reporting 30–50% savings against routing everything to Opus 4.8. Cursor published its routing rules but did not name a specific model per task type.","And underneath both, at the aggregation layer, OpenRouter warns that some providers serve quantized weights at lower prices, that output can differ from what full-precision weights would have produced, and that your logs will not tell you this happened.","Three products. Three different answers to the question what is a model . Zero cases telling us which one the law will accept.","Model identity is fracturing along three independent axes, and they are not usually distinguished:","The engineering community treats all three as reliability problems. They are also identity problems, and identity is what contracts, warranties, disclosures, and evidence rules are built on.","Start with the oldest question in commercial law: was the description a term of the deal?","If an API call were a sale of goods, this would be near-trivial. UCC §2-313(1)(b) makes any description of the goods that forms part of the basis of the bargain into an express warranty that the goods will conform to it. India’s Sale of Goods Act, 1930, §15 does the same work through the doctrine of sale by description.","But an inference API almost certainly isn’t goods. Courts have generally treated hosted software as a service, which pushes you out of warranty statutes and into common-law contract — where the answer depends entirely on what the documentation said and how specifically the buyer bargained. That is precisely where the ambiguity lives. Enterprise agreements price per-model. Model cards are model-specific. Compliance artifacts name model versions. And then the routing layer treats the name as a hint .","The cleanest way to see the stakes: if your code pins a model ID rather than expressing a capability requirement, behaviour can shift materially without any error ever firing. A contract drafted the same way has the same defect. If you bargained for a name, substitution is breach. If you bargained for a capability, substitution is fine — and now you need a definition of “frontier quality” that survives cross-examination.","Nobody has written that definition. Cursor’s is instructive: Router was evaluated in an online A/B test optimising for user satisfaction as the reward signal. That is a sound engineering choice and a strange contractual one. Satisfaction is not conformity. A user who never noticed the swap is evidence of a good router, not of a delivered specification.","Under FTC deception doctrine, a representation is actionable when it is material and likely to mislead a consumer acting reasonably; objective performance claims additionally require a reasonable basis before dissemination. Both halves bite here.","On the routing side, the three products sit at very different points. Anthropic notifies and returns the served model in the response. Cursor publishes rules but not per-task model assignment. OpenRouter’s quantization variance is disclosed in documentation and controllable by parameter but it is opt-out, and the default path is the cheap one.","Disclosure buried in a docs page, defaulted against the user, is exactly the fact pattern regulators have been calling a dark pattern in every other consumer vertical.","On the substantiation side, the exposure is the claim , not the routing. Commentators have already flagged that a “60% cheaper, no quality loss” claim arrives without published methodology, and Anthropic’s own disclosure is a model of what candour costs: it stated plainly that the retrained classifier flags benign requests more often during routine coding and debugging. That sentence is a liability shield. The vendors who don’t write it are the ones to watch.","There is a competition-law tail here too. Working papers on vertical foreclosure in inference markets are already proposing a conduct framework built on routing transparency, quality-of-service parity, and FRAND-style non-discrimination. A router that is also a first-party model vendor is a self-preferencing engine wearing a cost-optimisation costume.","This is where I think the real fight lands, and it has nothing to do with billing. Federal Rule of Evidence 901(b)(9) authenticates output by describing the process or system that produced it and showing that the process produces an accurate result. FRE 902(13) and 902(14) go further, letting records generated by an electronic system, or data verified by hash, self-authenticate on certification. India’s Bharatiya Sakshya Adhiniyam, 2023, §63 does the analogous work, conditioning admissibility of an electronic record on a certificate identifying the record and the manner of its production.","Every one of these provisions assumes you can name the system.","Now run the scenario – A lawyer files a brief with a fabricated citation. Sanctions proceedings begin. The court asks which model produced it. The firm’s logs say fable-5. The provider’s logs say a classifier fired and Opus 4.8 answered. Or the request went through an IDE router that selected a model the firm cannot reconstruct, at a precision it never specified, on a version that has since been retired.","The chain of custody breaks at the router. Not at the model at the dispatcher . And it breaks in both directions: the party trying to authenticate its own output can’t, and the party trying to attack the other side’s output has a free-standing reliability argument that no expert can rebut.","Add the discovery problem. Routing decisions are now dispositive facts. That means router classifier logs, per-request model assignments, and precision metadata are discoverable from a vendor who currently has no obligation to retain them.","You cannot draft your way out of this, because the representation you need is about a runtime event, and contracts are static.","What the law will eventually ask for is attestable model identity : a signed assertion, returned with each response, binding the completion to a tuple of served model identifier, weights hash, precision, and system-prompt hash signed by a key rooted in hardware attestation, which confidential-computing GPU platforms already support.","The primitive is half-built. Response objects already carry the served model name. What is missing is the property that makes it legally useful: that it cannot be asserted falsely, and that the buyer can verify it without trusting the seller. A hash in a response header does for model identity exactly what FRE 902(14) already does for copied data turns a contested factual question into a certificate.","Routing is good engineering. It is also, right now, an unlogged, unsigned, unverifiable substitution of the thing you bargained for. The model ID has quietly become a legal identifier. It is time somebody decided what it identifies.","Aabis Islam is a student pursuing a BA LLB at National Law University, Delhi. With a strong interest in AI Law, Aabis is passionate about exploring the intersection of artificial intelligence and legal frameworks. Dedicated to understanding the implications of AI in various legal contexts, Aabis is keen on investigating the advancements in AI technologies and their practical applications in the legal field."],"articleImages":[{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2024/06/IMG_2925-Aabis-Islam-300x300.png","alt":"","afterParagraph":26,"url":"/media/articles/cmrxueg0q02ryroxpw82u6uld/ff977ed957be7457.png"},{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2026/07/blog6171-1-100x70.png","alt":"Designing High-Performance GPU Kernels with TileLang: Tensor-Core GEMM, Fused Softmax, FlashAttention, and Autotuning","afterParagraph":27,"url":"/media/articles/cmrxueg0q02ryroxpw82u6uld/80925b48ffa65ad2.webp"}],"mediaStatus":"ok","articleBodyZh":["你调用了 API。你传递了 model: “claude-fable-5”。你收到了一个完成结果、一个令牌计数，以及一个字段显示 “model”: “claude-opus-4-8”。","没有出现错误，也没有重试。请求在生成开始前就已被分类，匹配了一个敏感类别，并被完全交给另一组权重处理。Anthropic 在 7 月 1 日重新发布 Fable 5 时记录了这一点：被阻止的请求会转发到 Opus 4.8，用户会收到通知。切换发生在 API 层，响应对象显示实际运行的模型。","那是表现正常的版本。据我所知，它也是唯一一个在广泛部署中会在数据流中告诉你真相的版本。","两周后，Cursor 发布了 Router：一个在 60 万以上的实时请求上训练的分类器，它会读取每个查询的上下文、复杂性和领域，并将其分派给它认为最合适的模型，三个早期访问账户报告与将所有请求都发送到 Opus 4.8 相比，可节省 30–50%。Cursor 公布了其路由规则，但没有为每种任务类型指定具体模型。","在两者之下的聚合层，OpenRouter 警告说，一些提供商以更低价格提供量化权重，输出可能与全精度权重生成的结果不同，而且你的日志不会告诉你这种情况发生了。","三个产品。对于“什么是模型”这个问题，三个不同的答案。没有任何案例告诉我们法律会接受哪一个。","模型身份沿三个独立轴线分化，并且通常不会区分：","工程社区将三者都视为可靠性问题。它们也是身份问题，而身份是合同、保修、披露和证据规则建立的基础。","从商业法中最古老的问题开始：描述是否构成交易条款？","如果 API 调用被视为货物销售，这几乎是微不足道的。UCC 第 2-313(1)(b) 条规定，任何构成交易基础的商品描述都成为明示保证，保证商品符合描述。印度 1930 年《货物销售法》第 15 条通过“按描述销售”的原则完成同样的功能。","但推理 API 几乎肯定不是商品。法院通常将托管的软件视为一种服务，这会将你排除在保修法规之外，而转入普通法合同——在这里，答案完全取决于文档内容以及买方具体如何谈判。这正是模糊性的所在。企业协议按模型定价。模型卡是针对特定模型的。合规性文件注明模型版本。然后，路由层将名称视为提示。","最清晰地看到利害关系的方法是：如果你的代码固定了一个模型 ID，而不是表达能力要求，那么行为可能会发生重大变化，而不会触发任何错误。以相同方式起草的合同也有同样的缺陷。如果你谈判的是名称，替换就是违约。如果你谈判的是能力，替换是可以接受的——现在你需要一个“前沿质量”的定义，该定义能够经得起交叉盘问。","目前没人写过这个定义。Cursor 的做法具有启发性：Router 在在线 A/B 测试中进行了评估，以用户满意度作为奖励信号优化。这是一个合理的工程选择，但合同上却很奇怪。满意度不等于符合规范。一个从未注意到替换的用户，是一个优秀路由器的证据，而不是已交付规格的证据。","根据 FTC 欺诈行为法，若陈述具重大性且可能误导合理行为的消费者，则可采取法律行动；客观性能宣称在传播之前还需有合理依据。这里两方面都适用。","在路由方面，这三款产品的位置大不相同。Anthropic 会在响应中通知并返回所提供的模型。Cursor 仅发布规则，但不进行按任务模型分配。OpenRouter 的量化差异在文档中披露，并可通过参数控制，但这是选择退出的方式，默认路径是廉价的。","埋藏在文档页面中的披露、且默认对用户不利，正是监管机构在其他消费领域称为“灰暗模式”的事实模式。","在论证方面，风险点是声明本身，而不是路由。评论者已经指出，一个“便宜60%，质量无损”的声明出现时，没有提供公开的方法论，而Anthropic自己的披露展示了坦白的代价：它明确表示，重新训练后的分类器在日常编码和调试过程中更频繁地标记良性请求。那句话成为了一种责任屏障。那些没有写明的供应商才是值得关注的对象。","这里也涉及竞争法问题。有关推理市场垂直排他行为的工作论文已经提出了一个基于路由透明度、服务质量平价和FRAND式非歧视的行为框架。作为第一方模型供应商的路由器，是一种披着成本优化外衣的自我偏好引擎。","我认为真正的争论点在这里，这与计费无关。联邦证据规则901(b)(9)通过描述生成输出的过程或系统，并证明该过程能产生准确结果来验证输出的真实性。FRE 902(13)和902(14)更进一步，允许由电子系统生成的记录或通过哈希验证的数据在获得认证时自我证明。印度《2023年印度电子证据法》第63条也进行了类似规定，将电子记录的可采性与识别该记录及其生成方式的证书挂钩。","这些条款都假设你能命名该系统。","现在运行这个情景——一名律师提交了含有虚构引用的诉状。开始进行制裁程序。法院询问是哪个模型生成的。律所的日志显示是fable-5。供应商的日志显示触发了一次分类器，并由Opus 4.8作答。或者请求通过了一个IDE路由器，选中的模型是律所无法重建的，其精度从未被指定，并且版本已经退休。","监管链在路由器处中断，而不是在调度器的模型处中断。并且这种中断是双向的：尝试验证自身输出的一方无法完成验证，而试图攻击对方输出的一方拥有一个任何专家都无法驳斥的独立可靠性论点。","增加发现问题。路由决策现在是决定性事实。这意味着路由器分类器日志、每次请求的模型分配以及精度元数据可以从当前没有义务保留这些数据的供应商处获取。","你不能通过起草合同来解决这个问题，因为你需要的表示形式是关于运行时事件的，而合同是静态的。","法律最终会要求的是可证明的模型身份：每次响应返回时附带的签名声明，将完成结果绑定到一个包含服务模型标识符、权重哈希、精度以及由硬件认证根密钥签署的系统提示哈希的元组上，这一点是机密计算 GPU 平台已经支持的。","这一原语已经半成。响应对象已经携带了服务模型的名称。缺失的是使其在法律上有用的属性：即它不能被虚假声明，并且购买者可以无需信任卖方就能验证它。响应头中的哈希对模型身份的作用，正如 FRE 902(14) 对复制数据所起的作用一样，能够将有争议的事实问题转化为证书。","路由是良好的工程实践。但现在，它也是一个未记录、未签名、无法验证的对你所谈条件物的替代品。模型 ID 已悄然成为法律标识符。是时候有人决定它究竟标识什么了。","Aabis Islam 是德里国家法律大学的 BA LLB 学生。Aabis 对人工智能法律有浓厚兴趣，对探索人工智能与法律框架的交叉点充满热情。Aabis 致力于理解人工智能在各种法律背景下的影响，并热衷于研究人工智能技术的进展及其在法律领域的实际应用。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：调用 API 时指定 `model： \"claude-fable-5\"`，返回的响应中 `model` 字段却显示 `\"claude-opus-4-8\"`。 Aioga 将其归入「技巧观点」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":"2026-07-28T06:29:10.583Z","sourceHash":"55f531b6a04cf537","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["技巧观点","MarkTechPost（RSS）"],"translations":{"zh-CN":{"title":"你调用的 AI 模型可能不是你付钱的那个--Anthropic 被曝静默切换模型","summary":"调用 API 时指定 `model： \"claude-fable-5\"`，返回的响应中 `model` 字段却显示 `\"claude-opus-4-8\"`。Anthropic 在生成开始前对请求进行分类，匹配到敏感类别后静默切换至另一组权重，且不报错、不重试。用户实际获得的模型与付费指定的模型不一致。","category":"技巧观点","source":"marktechpost.com","aggregationSource":"MarkTechPost（RSS）","pageTitle":"你调用的 AI 模型可能不是你付钱的那个--Anthropic 被曝静默切换模型 - Aioga AI资讯","description":"调用 API 时指定 `model： \"claude-fable-5\"`，返回的响应中 `model` 字段却显示 `\"claude-opus-4-8\"`。Anthropic 在生成开始前对请求进行分类，匹配到敏感类别后静默切换至另一组权重，且不报错、不重试。用户实际获得的模型与付费指定的模型不一致。","url":"https://www.aioga.com/news/cmrxueg0q02ryroxpw82u6uld/"},"en":{"title":"The AI model you called may not be the one you paid for--Anthropic was exposed for silently switching models","summary":"When specifying `model: \"claude-fable-5\"` while calling the API, the `model` field in the returned response shows `\"claude-opus-4-8\"`. Before generation, Anthropic classifies the request, and if it matches a sensitive category, it silently switches to another set of weights without reporting an error or retrying. The model actually received by the user is not the same as the model paid for.","category":"Insights","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"The AI model you called may not be the one you paid for--Anthropic was exposed for silently switching models - Aioga AI News","description":"When specifying `model: \"claude-fable-5\"` while calling the API, the `model` field in the returned response shows `\"claude-opus-4-8\"`. Before generation, Anthropic classifies the r...","url":"https://www.aioga.com/en/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:42:09.919Z"},"ja":{"title":"あなたが呼び出しているAIモデルは、あなたが支払ったものではない可能性があります――Anthropicがモデルを静かに切り替えたと報じられる","summary":"APIを呼び出す際には「model: \"claude-fable-5\"と指定しますが、返送されるレスポンスでは「model」フィールドが「claude-opus-4-8」と表示されます。 生成が始まる前に、Anthropicはリクエストを分類し、センシティブなカテゴリを一致させた後、エラーや再試行なく静かに別の重み付けに切り替えます。 ユーザーが実際に受け取るモデルは、支払いに指定されたモデルと一致しません。","category":"ヒントと視点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"あなたが呼び出しているAIモデルは、あなたが支払ったものではない可能性があります――Anthropicがモデルを静かに切り替えたと報じられる - Aioga AIニュース","description":"APIを呼び出す際には「model: \"claude-fable-5\"と指定しますが、返送されるレスポンスでは「model」フィールドが「claude-opus-4-8」と表示されます。 生成が始まる前に、Anthropicはリクエストを分類し、センシティブなカテゴリを一致させた後、エラーや再試行なく静かに別の重み付けに切り替えます。 ユーザーが実際に受け取...","url":"https://www.aioga.com/ja/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:42:23.397Z"},"ko":{"title":"당신이 호출한 AI 모델은 당신이 돈을 지불한 것이 아닐 수도 있습니다--Anthropic이 조용히 모델을 전환한 것으로 드러남","summary":"API를 호출할 때 'model: \"claude-fable-5\"를 지정하지만, 반환된 응답에서는 'model' 필드가 'claude-opus-4-8''로 표시됩니다. 생성이 시작되기 전에 Anthropic은 요청을 분류하고, 민감한 카테고리를 매칭한 후 오류나 재시도 없이 조용히 다른 가중치 집합으로 전환합니다. 사용자가 실제로 받는 모델은 지불에 지정된 모델과 일치하지 않습니다.","category":"인사이트","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"당신이 호출한 AI 모델은 당신이 돈을 지불한 것이 아닐 수도 있습니다--Anthropic이 조용히 모델을 전환한 것으로 드러남 - Aioga AI 뉴스","description":"API를 호출할 때 'model: \"claude-fable-5\"를 지정하지만, 반환된 응답에서는 'model' 필드가 'claude-opus-4-8''로 표시됩니다. 생성이 시작되기 전에 Anthropic은 요청을 분류하고, 민감한 카테고리를 매칭한 후 오류나 재시도 없이 조용히 다른 가중치 집합으로 전환합니다. 사용자가...","url":"https://www.aioga.com/ko/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:43:19.516Z"},"es":{"title":"El modelo de IA que estás utilizando puede no ser el por el que pagaste: se revela que Anthropic cambió silenciosamente el modelo","summary":"Al llamar a la API, especifica 'model: \"claude-fable-5\"', pero la respuesta devuelto muestra el campo 'model' como 'claude-opus-4-8\"'. Antes de que comience la generación, Anthropic clasifica las solicitudes y, tras emparejar categorías sensibles, cambia silenciosamente a otro conjunto de pesos sin errores ni repeticiones. El modelo que realmente reciben los usuarios no coincide con el modelo especificado para el pago.","category":"Ideas","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"El modelo de IA que estás utilizando puede no ser el por el que pagaste: se revela que Anthropic cambió silenciosamente el modelo - Aioga Noticias de IA","description":"Al llamar a la API, especifica 'model: \"claude-fable-5\"', pero la respuesta devuelto muestra el campo 'model' como 'claude-opus-4-8\"'. Antes de que comience la generación, Anthropi...","url":"https://www.aioga.com/es/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:43:11.495Z"},"fr":{"title":"Le modèle d'IA que vous utilisez pourrait ne pas être celui pour lequel vous avez payé - Anthropic aurait discrètement changé de modèle","summary":"Lorsque vous appelez l'API en spécifiant `model : \"claude-fable-5\"`, le champ `model` dans la réponse renvoyée affiche néanmoins \"claude-opus-4-8\". Anthropic classe la requête avant la génération, et si elle correspond à une catégorie sensible, elle bascule silencieusement vers un autre ensemble de poids, sans générer d'erreur ni réessayer. Le modèle réellement obtenu par l'utilisateur n'est pas celui qu'il a payé pour spécifier.","category":"Analyses","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Le modèle d'IA que vous utilisez pourrait ne pas être celui pour lequel vous avez payé - Anthropic aurait discrètement changé de modèle - Aioga Actualités IA","description":"Lorsque vous appelez l'API en spécifiant `model : \"claude-fable-5\"`, le champ `model` dans la réponse renvoyée affiche néanmoins \"claude-opus-4-8\". Anthropic classe la requête avan...","url":"https://www.aioga.com/fr/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:44:03.370Z"},"de":{"title":"Das von Ihnen aufgerufene KI-Modell ist möglicherweise nicht dasjenige, für das Sie bezahlt haben – Anthropic wurde beschuldigt, das Modell stillschweigend zu wechseln","summary":"Beim Aufruf der API wird `model: \"claude-fable-5\"` angegeben, aber im zurückgegebenen Antwortfeld `model` wird \"claude-opus-4-8\" angezeigt. Anthropic klassifiziert die Anfragen vor der Generierung und wechselt stillschweigend zu einem anderen Gewichtsset, wenn eine Anfrage in eine sensible Kategorie fällt, ohne einen Fehler auszugeben oder einen erneuten Versuch zu starten. Der Benutzer erhält tatsächlich ein anderes Modell als das bezahlte, angegebene Modell.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Das von Ihnen aufgerufene KI-Modell ist möglicherweise nicht dasjenige, für das Sie bezahlt haben – Anthropic wurde beschuldigt, das Modell stillschweigend zu wechseln - Aioga KI-News","description":"Beim Aufruf der API wird `model: \"claude-fable-5\"` angegeben, aber im zurückgegebenen Antwortfeld `model` wird \"claude-opus-4-8\" angezeigt. Anthropic klassifiziert die Anfragen vor...","url":"https://www.aioga.com/de/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:44:07.914Z"},"pt-BR":{"title":"O modelo de IA que você está usando pode não ser aquele pelo qual você pagou – Anthropic foi exposto por trocar silenciosamente o modelo","summary":"Ao chamar a API especificando `model: \"claude-fable-5\"`, o campo `model` na resposta retornada mostra, no entanto, \"claude-opus-4-8\". A Anthropic classifica a solicitação antes de iniciar a geração e, se corresponder a uma categoria sensível, altera silenciosamente para outro conjunto de pesos, sem gerar erro ou tentar novamente. O modelo que o usuário obtém na prática não corresponde ao modelo especificado e pago.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"O modelo de IA que você está usando pode não ser aquele pelo qual você pagou – Anthropic foi exposto por trocar silenciosamente o modelo - Aioga Notícias de IA","description":"Ao chamar a API especificando `model: \"claude-fable-5\"`, o campo `model` na resposta retornada mostra, no entanto, \"claude-opus-4-8\". A Anthropic classifica a solicitação antes de...","url":"https://www.aioga.com/pt-BR/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:44:52.481Z"},"ru":{"title":"AI-модель, которую вы вызываете, возможно, не та, за которую вы платите — Anthropic был замечен в тихом переключении модели","summary":"При вызове API с указанием `model: \"claude-fable-5\"`, в возвращаемом ответе поле `model` отображается как `\"claude-opus-4-8\"`. Anthropic классифицирует запрос перед началом генерации, и если он попадает в чувствительную категорию, автоматически переключается на другой набор весов без ошибок и повторных попыток. Пользователь фактически получает модель, которая не соответствует оплаченной и указанной модели.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"AI-модель, которую вы вызываете, возможно, не та, за которую вы платите — Anthropic был замечен в тихом переключении модели - Aioga Новости ИИ","description":"При вызове API с указанием `model: \"claude-fable-5\"`, в возвращаемом ответе поле `model` отображается как `\"claude-opus-4-8\"`. Anthropic классифицирует запрос перед началом генерац...","url":"https://www.aioga.com/ru/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:44:57.025Z"},"ar":{"title":"النموذج الذكي الاصطناعي الذي تستخدمه قد لا يكون الذي دفعت مقابله -- تم الكشف عن تغيير نموذجي صامت من قبل Anthropic","summary":"عند استدعاء واجهة برمجة التطبيقات، حدد 'model: \"claude-fable-5\"'، لكن الرد المردود يظهر حقل 'model' ك '\"claude-opus-4-8\"'. قبل بدء التوليد، يقوم Anthropic بتصنيف الطلبات، وبعد مطابقة الفئات الحساسة، ينتقل بصمت إلى مجموعة أوزان أخرى دون أخطاء أو إعادة محاولات. النموذج الذي يتلقاه المستخدمون فعليا لا يتطابق مع النموذج المحدد للدفع.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"النموذج الذكي الاصطناعي الذي تستخدمه قد لا يكون الذي دفعت مقابله -- تم الكشف عن تغيير نموذجي صامت من قبل Anthropic - Aioga أخبار الذكاء الاصطناعي","description":"عند استدعاء واجهة برمجة التطبيقات، حدد 'model: \"claude-fable-5\"'، لكن الرد المردود يظهر حقل 'model' ك '\"claude-opus-4-8\"'. قبل بدء التوليد، يقوم Anthropic بتصنيف الطلبات، وبعد مطاب...","url":"https://www.aioga.com/ar/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:45:43.517Z"},"hi":{"title":"आप जो AI मॉडल कॉल कर रहे हैं वह शायद वह नहीं है जिसका आप भुगतान कर रहे हैं--Anthropic पर मौन में मॉडल बदलने का आरोप","summary":"एपीआई को कॉल करते समय, 'मॉडल: \"क्लाउड-फैबल-5'' निर्दिष्ट करें, लेकिन लौटाया गया जवाब 'मॉडल' फ़ील्ड को ''क्लाउड-ओपस-4-8'' के रूप में दिखाता है। पीढ़ी शुरू होने से पहले, एंथ्रोपिक अनुरोधों को वर्गीकृत करता है, और संवेदनशील श्रेणियों का मिलान करने के बाद, चुपचाप त्रुटियों या पुन: प्रयासों के बिना वजन के दूसरे सेट पर स्विच करता है। उपयोगकर्ताओं को वास्तव में प्राप्त होने वाला मॉडल भुगतान के लिए निर्दिष्ट मॉडल से मेल नहीं खाता है।","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"आप जो AI मॉडल कॉल कर रहे हैं वह शायद वह नहीं है जिसका आप भुगतान कर रहे हैं--Anthropic पर मौन में मॉडल बदलने का आरोप - Aioga AI समाचार","description":"एपीआई को कॉल करते समय, 'मॉडल: \"क्लाउड-फैबल-5'' निर्दिष्ट करें, लेकिन लौटाया गया जवाब 'मॉडल' फ़ील्ड को ''क्लाउड-ओपस-4-8'' के रूप में दिखाता है। पीढ़ी शुरू होने से पहले, एंथ्रोपिक अन...","url":"https://www.aioga.com/hi/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:45:57.504Z"},"it":{"title":"Il modello AI che hai chiamato potrebbe non essere quello per cui hai pagato: Anthropic è stato accusato di cambiare silenziosamente il modello","summary":"Quando si chiama l'API specificando `model: \"claude-fable-5\"`, nel campo `model` della risposta viene invece mostrato `\"claude-opus-4-8\"`. Anthropic classifica la richiesta prima di generare la risposta e, se corrisponde a una categoria sensibile, passa silenziosamente a un altro set di pesi, senza generare errori e senza riprovare. L'utente ottiene quindi effettivamente un modello diverso da quello specificato a pagamento.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Il modello AI che hai chiamato potrebbe non essere quello per cui hai pagato: Anthropic è stato accusato di cambiare silenziosamente il modello - Aioga Notizie IA","description":"Quando si chiama l'API specificando `model: \"claude-fable-5\"`, nel campo `model` della risposta viene invece mostrato `\"claude-opus-4-8\"`. Anthropic classifica la richiesta prima d...","url":"https://www.aioga.com/it/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:46:48.180Z"},"nl":{"title":"Het AI-model dat je oproept, is mogelijk niet het model waarvoor je betaalt—Anthropic werd beschuldigd van stilzwijgende modelwissel","summary":"Bij het aanroepen van de API specificeer je 'model: \"claude-fable-5\"', maar het teruggegeven antwoord toont het 'model'-veld als 'claude-opus-4-8''. Voordat de generatie begint, classificeert Anthropic verzoeken en schakelt na het matchen van gevoelige categorieën stilletjes over naar een andere set gewichten zonder fouten of herpogingen. Het model dat gebruikers daadwerkelijk ontvangen, komt niet overeen met het model dat voor betaling is gespecificeerd.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Het AI-model dat je oproept, is mogelijk niet het model waarvoor je betaalt—Anthropic werd beschuldigd van stilzwijgende modelwissel - Aioga AI-nieuws","description":"Bij het aanroepen van de API specificeer je 'model: \"claude-fable-5\"', maar het teruggegeven antwoord toont het 'model'-veld als 'claude-opus-4-8''. Voordat de generatie begint, cl...","url":"https://www.aioga.com/nl/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:46:46.269Z"},"tr":{"title":"Kullandığınız AI modeli, ödediğiniz model olmayabilir--Anthropic sessizce modeli değiştirdiği ortaya çıktı","summary":"API'yi çağırırken, 'model: \"claude-fable-5\"' belirtin, ancak geri dönen yanıt 'model' alanını 'claude-opus-4-8'' olarak gösterir. Üretim başlamadan önce, Anthropic istekler sınıflandırır ve hassas kategorileri eşleştirdikten sonra hata veya tekrar denemeler olmadan sessizce başka bir ağırlık setine geçer. Kullanıcıların aldığı model, ödeme için belirtilen modelle eşleşmiyor.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Kullandığınız AI modeli, ödediğiniz model olmayabilir--Anthropic sessizce modeli değiştirdiği ortaya çıktı - Aioga AI Haberleri","description":"API'yi çağırırken, 'model: \"claude-fable-5\"' belirtin, ancak geri dönen yanıt 'model' alanını 'claude-opus-4-8'' olarak gösterir. Üretim başlamadan önce, Anthropic istekler sınıfla...","url":"https://www.aioga.com/tr/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:47:33.813Z"},"vi":{"title":"Mô hình AI mà bạn gọi có thể không phải là mô hình mà bạn đã trả tiền -- Anthropic bị tiết lộ đã âm thầm chuyển đổi mô hình","summary":"Khi gọi API chỉ định `model: \"claude-fable-5\"`, trong phản hồi trả về trường `model` lại hiển thị \"claude-opus-4-8\". Anthropic phân loại yêu cầu trước khi bắt đầu sinh, nếu khớp với danh mục nhạy cảm thì sẽ lặng lẽ chuyển sang một bộ trọng số khác, và không báo lỗi, không thử lại. Mô hình mà người dùng thực sự nhận được không khớp với mô hình đã chỉ định khi trả tiền.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Mô hình AI mà bạn gọi có thể không phải là mô hình mà bạn đã trả tiền -- Anthropic bị tiết lộ đã âm thầm chuyển đổi mô hình - Tin tức AI Aioga","description":"Khi gọi API chỉ định `model: \"claude-fable-5\"`, trong phản hồi trả về trường `model` lại hiển thị \"claude-opus-4-8\". Anthropic phân loại yêu cầu trước khi bắt đầu sinh, nếu khớp vớ...","url":"https://www.aioga.com/vi/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:47:37.367Z"},"id":{"title":"Model AI yang Anda panggil mungkin bukan yang Anda bayar--Anthropic dilaporkan diam-diam mengganti model","summary":"Saat memanggil API dengan menentukan `model: \"claude-fable-5\"`, bidang `model` dalam respons yang dikembalikan justru menampilkan `\"claude-opus-4-8\"`. Anthropic mengklasifikasikan permintaan sebelum mulai menghasilkan, dan jika cocok dengan kategori sensitif, secara diam-diam beralih ke set bobot lain tanpa menampilkan kesalahan atau melakukan percobaan ulang. Model yang sebenarnya diperoleh pengguna tidak sesuai dengan model yang ditentukan untuk pembayaran.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Model AI yang Anda panggil mungkin bukan yang Anda bayar--Anthropic dilaporkan diam-diam mengganti model - Berita AI Aioga","description":"Saat memanggil API dengan menentukan `model: \"claude-fable-5\"`, bidang `model` dalam respons yang dikembalikan justru menampilkan `\"claude-opus-4-8\"`. Anthropic mengklasifikasikan...","url":"https://www.aioga.com/id/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:48:20.843Z"},"th":{"title":"โมเดล AI ที่คุณเรียกใช้อาจไม่ใช่โมเดลที่คุณจ่ายเงินไป -- Anthropic ถูกเปิดเผยว่ามีการเปลี่ยนโมเดลอย่างเงียบ ๆ","summary":"เมื่อเรียกใช้ API และระบุ `model: \"claude-fable-5\"` ช่อง `model` ในการตอบกลับกลับแสดงเป็น `\"claude-opus-4-8\"` Anthropic จะทำการจัดประเภทคำขอก่อนเริ่มสร้าง และหากตรงกับหมวดหมู่ที่ละเอียดอ่อน จะทำการเปลี่ยนน้ำหนักไปยังชุดอื่นอย่างเงียบ ๆ โดยไม่รายงานข้อผิดพลาดหรือทำการลองใหม่ ผู้ใช้จะได้โมเดลจริงที่ใช้ไม่ตรงกับโมเดลที่จ่ายเงินระบุ","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"โมเดล AI ที่คุณเรียกใช้อาจไม่ใช่โมเดลที่คุณจ่ายเงินไป -- Anthropic ถูกเปิดเผยว่ามีการเปลี่ยนโมเดลอย่างเงียบ ๆ - ข่าว AI Aioga","description":"เมื่อเรียกใช้ API และระบุ `model: \"claude-fable-5\"` ช่อง `model` ในการตอบกลับกลับแสดงเป็น `\"claude-opus-4-8\"` Anthropic จะทำการจัดประเภทคำขอก่อนเริ่มสร้าง และหากตรงกับหมวดหมู่ที่ละ...","url":"https://www.aioga.com/th/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:48:32.445Z"},"pl":{"title":"Model AI, którego używasz, może nie być tym, za który zapłaciłeś – Anthropic został ujawniony, że cicho zmienił model","summary":"Wywołując API, określ 'model: \"claude-fable-5\"', ale zwrócona odpowiedź pokazuje pole 'model' jako '\"claude-opus-4-8\"'. Przed rozpoczęciem generowania Anthropic klasyfikuje żądania, a po dopasowaniu wrażliwych kategorii cicho przełącza się na inny zestaw wag bez błędów i powtórek. Model, który faktycznie otrzymują użytkownicy, nie odpowiada modelu określonemu do płatności.","category":"技巧观点","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Model AI, którego używasz, może nie być tym, za który zapłaciłeś – Anthropic został ujawniony, że cicho zmienił model - Aioga Wiadomości AI","description":"Wywołując API, określ 'model: \"claude-fable-5\"', ale zwrócona odpowiedź pokazuje pole 'model' jako '\"claude-opus-4-8\"'. Przed rozpoczęciem generowania Anthropic klasyfikuje żądania...","url":"https://www.aioga.com/pl/news/cmrxueg0q02ryroxpw82u6uld/","contentTranslated":true,"sourceHash":"a410dc5fd557cb56","translatedAt":"2026-07-27T06:49:21.335Z"}}}}