Show me top 10 trending question Chinese Mainland users often ask about Claude Opus 4.7 now. Show me both Chinese Mainland (zh CN) language & English version for me to understand these questions.
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent perform...
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and latency, and the workflow changes needed to use it well. I can’t verify exact Main
要点
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and la
I can’t verify exact Mainland search volume rankings from open sources, so this is a best effort trend synthesis rather than an official leaderboard.
I’ve written each question in simplified Chinese and English so you can scan them quickly.
[13][16] Key findings 1.
zh CN：大陆现在能直接注册并稳定使用 Claude Opus 4.7 吗？ English: Can users in Mainland China register for and use Claude Opus 4.7 reliably right now?
zh CN：如果我现在用的是 Opus 4.6，升级到 4.7 到底值不值？ English: If I’m using Opus 4.6 now, is upgrading to 4.7 actually worth it?
研究答案
Based on currently indexed Chinese-language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and latency, and the workflow changes needed to use it well. I can’t verify exact Mainland search-volume rankings from open sources, so this is a best-effort trend synthesis rather than an official leaderboard. I’ve written each question in simplified Chinese and English so you can scan them quickly. [13][16]
Key findings
zh-CN：大陆现在能直接注册并稳定使用 Claude Opus 4.7 吗？
English: Can users in Mainland China register for and use Claude Opus 4.7 reliably right now? [8][11]
zh-CN：如果我现在用的是 Opus 4.6，升级到 4.7 到底值不值？
English: If I’m using Opus 4.6 now, is upgrading to 4.7 actually worth it? [12][13][16]
zh-CN：4.7 在编程、前端开发和长任务 Agent 上到底强多少？
English: How much better is 4.7 for coding, frontend work, and long-running agent tasks? [12][16]
zh-CN：为什么很多人说 4.7 更会“接情绪”了，但写作风格反而变差了？
English: Why do many users say 4.7 feels more emotionally accommodating, but worse in writing style? [16]
zh-CN：官方价格看起来没明显涨，为什么实际感觉更费 tokens、更慢、更贵？
English: If the official pricing doesn’t look much higher, why does it feel more token-hungry, slower, and costlier in practice? [10][13][16]
zh-CN：在 Claude Code 或 API 里，adaptive thinking、effort、xhigh 应该怎么设置？
English: In Claude Code or the API, how should adaptive thinking, effort, and xhigh be configured? [7][13]
zh-CN：1M 上下文窗口到底有没有实际价值，最适合哪些场景？
English: How useful is the 1M-context window in real work, and which use cases benefit most? [9][12]
zh-CN：4.7 的视觉能力到底提升了多少，能不能更稳地看 UI、截图、图表和 PDF？
English: How much did 4.7’s vision improve, and is it more reliable on UIs, screenshots, charts, and PDFs? [12][16]
zh-CN：从 4.6 迁移到 4.7，会踩到哪些 API、参数或提示词层面的坑？
English: What API, parameter, or prompting pitfalls appear when migrating from 4.6 to 4.7? [13]
zh-CN：4.7 的安全限制是不是更严了，会不会更容易拒答或拦截高风险请求？
English: Are 4.7’s safety restrictions stricter, and will it refuse or block high-risk requests more often? [13]
Confirmed facts
Official Claude docs say users should check supported regions for API availability, so location and access are a real practical concern behind many user questions. [8][11]
Official docs say Claude Opus 4.7 has a 1M-token context window and present it as improved on knowledge work, vision, and long-context tasks. [9][12]
Official migration docs say Opus 4.7 shifts to adaptive thinking, changes tokenization, removes some prior parameter behavior, and tends to use tools less by default. [7][13]
Official pricing docs list Opus 4.7 pricing alongside Opus 4.6, while Chinese launch coverage says the pricing strategy stayed the same as the prior version. [10][16]
Chinese-language coverage around launch highlights stronger coding and long-task reliability, but also surfaces complaints about writing tone, search feel, and overall “personality.” [16]
What remains inference
This is not a true search-volume ranking from Baidu, WeChat, Xiaohongshu, or internal Anthropic analytics.
The ordering is inferred from repeated themes in indexed Chinese-language coverage and developer discussion, which likely overrepresents technical users. [13][16]
What the evidence suggests
The biggest attention split is between access questions and “is 4.7 actually better for my workflow than 4.6?” questions. [8][11][13][16]
Developer-heavy questions dominate because the official changes are workflow-heavy: adaptive thinking, effort tuning, long context, migration changes, and tool-use behavior. [7][9][12][13]
The curiosity is not just about benchmark gains. People also care about whether daily use feels worse in style, speed, or cost. [13][16]
Conflicting evidence or uncertainty
Official docs frame 4.7 as stronger on coding, knowledge work, vision, and long-context use. [12]
Chinese-language commentary still reports mixed subjective reactions, especially around writing tone, search quality, and whether the model feels less distinctive. [16]
So the safest conclusion is that 4.7 looks especially compelling for coding and agentic workflows, but not everyone sees it as a universal upgrade for writing-heavy use. [12][16]
Open questions
How much of the negative reaction comes from the base model versus app-layer behavior, safety layers, or prompting habits? [13][16]
How are Mainland users actually accessing it in practice: direct account use, partner clouds, or other workflows? [8][11]
Is the perceived writing-style regression fixable with prompting and effort tuning, or is it a real tradeoff in this release? [7][13][16]
Sources worth trusting most
Official Claude docs on what changed in Opus 4.7, migration behavior, context windows, pricing, and supported regions. [7][9][10][11][12][13]
Chinese-language launch and reaction coverage for what users are actively debating right now. [16]
Recommended next step
If you want, I can next turn these 10 questions into a bilingual FAQ with short, factual answers in zh-CN + English, so you can use it for research, content planning, customer support, or market monitoring.
Summary
Your best current “top 10” for Mainland-facing Chinese discussion is: access, upgrade value, coding gains, writing-style complaints, token cost, adaptive-thinking setup, 1M context usefulness, vision gains, migration pitfalls, and stricter safety behavior. That pattern is well supported by the combination of official docs and current Chinese-language discussion, but the exact rank order is still an inference rather than hard search-volume proof. [8][9][12][13][16]
人们还问
“Show me top 10 trending question Chinese Mainland users often ask about Claude Opus 4.7 now. Show me both Chinese Mainland (zh CN) language & English version for me to understand these questions.”的简短答案是什么？
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and la
首先要验证的关键点是什么？
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and la I can’t verify exact Mainland search volume rankings from open sources, so this is a best effort trend synthesis rather than an official leaderboard.
接下来在实践中我应该做什么？
I’ve written each question in simplified Chinese and English so you can scan them quickly.
Structure document content and metadata with XML tags: When using multiple documents, wrap each document in i.j4i.i2
<document>
tags with i.j4i.i2
<document_content>
and i.j4i.i2
<source>
(and other metadata) subtags for clarity. i.j4i.i2
<document>
i.j4i.i2
<document_content>
i.j4i.i2
<source>
### Example multi-document structure Ground responses in quotes: For long document tasks, ask Claude to quote relevant parts of the documents first before carrying out its task. This helps Claude cut through the noise of the rest of the document's contents. ### Example quote extraction ### Model self-knowledge If you would like Claude to ident…
跳至主要內容 Google Cloud Documentation 控制台 English Deutsch Español Español – América Latina Français Indonesia Italiano Português Português – Brasil עברית 中文 – 简体 中文 – 繁體 日本語 한국어 登入 。 ,,["上次更新時間：2026-04-18 (世界標準時間)。"],[],[]]
Claude Opus 4.7 Text Generation • Anthropic • Proxied Claude Opus 4.7 is Anthropic's most capable generally available model to date. It is highly autonomous and performs exceptionally well on long-horizon agentic work, knowledge work, vision tasks, and memory tasks. | Model Info | | Context Window ↗ | 1,000,000 tokens | | Terms and License | link ↗ | | More information | link ↗ | ## Usage TypeScript ``` const response = await env.AI.run( 'anthropic/claude-opus-4.7', { messages: [ { role: 'user', content: 'What are the three laws of thermodynamics?', }, ], max_tokens: 1024, }, { gateway: { i…
February 12, 2026 Self-serve Enterprise plans Previously, Enterprise plans were only available to customers working with our Sales team. Now, any organization can purchase an Enterprise plan directly on our website with no Sales conversation required. Self-serve Enterprise plans have a single seat type that includes access to Claude, Claude Code, and Cowork. For more information, refer to our blog post or What is the Enterprise plan? ### February 5, 2026 Claude Opus 4.6 launch We’ve upgraded our smartest model and improved its coding skills. Read our blog post for more information: Introd…
Adaptive thinking is the recommended way to use extended thinking with Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6, and is the default mode on Claude Mythos Preview (where it auto-applies whenever i.j4i.i2
thinking
is unset). Instead of manually setting a thinking token budget, adaptive thinking lets Claude dynamically determine when and how much to use extended thinking based on the complexity of each request. On Claude Opus 4.7, adaptive thinking is the only supported thinking mode; manual i.j4i.i2
For detailed information about limits, tiers, and the token bucket algorithm used for rate limiting, see Rate limits. ### Availability The Claude API is available in many countries and regions worldwide. Check the supported regions page to confirm availability in your location. ## Next Steps Messages API reference Complete API specification for direct model interactionsClaude Managed Agents reference Agents, Sessions, and Environments endpointsClient SDKs Python, TypeScript, Java, Go, C#, Ruby, and PHPRate limits Usage tiers, spend limits, and token bucket algorithm Was this page helpful? [..…
Claude Mythos Preview, Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6 have a 1M-token context window. Other Claude models, including Claude Sonnet 4.5 and Sonnet 4 (deprecated), have a 200k-token context window. A single request can include up to 600 images or PDF pages (100 for models with a 200k-token context window). When sending many images or large documents, you may approach request size limits before the token limit. ## Context awareness in Claude Sonnet 4.6, Sonnet 4.5, and Haiku 4.5 [...] How it works: At the start of a conversation, Claude receives information about its tot…
Seychelles Sierra Leone Singapore Slovakia Slovenia Solomon Islands South Africa South Korea Spain Sri Lanka Suriname Sweden Switzerland Taiwan Tajikistan Tanzania Thailand Timor-Leste, Democratic Republic of Togo Tonga Trinidad and Tobago Tunisia Turkey Turkmenistan Tuvalu Uganda Ukraine (except Crimea, Donetsk, and Luhansk regions) United Arab Emirates United Kingdom United States of America Uruguay Uzbekistan Vanuatu Vietnam Zambia Zimbabwe [...] India Indonesia Iraq Ireland Israel Italy Jamaica Japan Jordan Kazakhstan Kenya Kiribati Kuwait Kyrgyzstan Laos Latvia Lebanon Lesotho Liberia Li…
parameters to give additional headroom, including compaction triggers. Claude Opus 4.7 provides a 1M context window at standard API pricing with no long-context premium. ## Capability improvements ### Knowledge work Claude Opus 4.7 shows meaningful gains on knowledge-worker tasks, particularly where the model needs to visually verify its own outputs: .docx redlining and .pptx editing — improved at producing and self-checking tracked changes and slide layouts. Charts and figure analysis — improved at programmatic tool-calling with image-processing librarie…
實時網路安全防護： Claude Opus 4.7 新增的功能，涉及禁止或高風險主題的請求可能導致拒絕。對於合法的安全工作，例如滲透測試、漏洞研究或紅隊，請申請 網路驗證計畫 以請求降低對網路內容的限制。有關背景，請參閱 防護、警告和上訴。 高解析度影像支援： Claude Opus 4.7 是第一個具有高解析度影像支援的 Claude 模型，長邊的最大影像解析度為 2576 像素（從先前模型的 1568 像素提高）。這為視覺密集型工作負載帶來了收益，對於電腦使用、螢幕截圖理解和文件分析特別有價值。高解析度支援是自動的，不需要測試版標頭或用戶端選擇加入。全解析度影像可以使用比先前模型多約 3 倍的影像代幣（每個影像最多 4,784 代幣，相比之前的大約 1,6…
We implemented several protective measures to combat prompt injection attacks, including specialized reinforcement learning training to help the model recognize and avoid these manipulations, and the deployment of detection systems that can halt the model’s execution when a potential injection attempt is identi fi ed. These end-to-end defenses improved both models' prompt injection safety scores compared to their performance without safeguards. (see table 3.2A). Model Attack prevention score (without safeguards) Attack prevention score (with safeguards) Claude Opus 4 71% 89% Claude Sonnet 4 6…
AsiaBAN AsiaBAN # Anthropic bans Chinese-controlled entities from accessing Claude models Credit: AFP US AI startup Anthropic has enforced a new policy prohibiting Chinese-controlled entities from using its Claude series models, prompting ByteDance's Singapore-based IDE platform Trae to immediately remove Claude from its offerings. The move has attracted... Keep me signed in Some subscribers prefer to save their log-in information so they do not have to enter their User ID and Password each time they visit the site. To activate this function, check the 'Keep me signed in' box in the log-in se…
Japan Kenya Kiribati Kuwait Lebanon Lesotho Liberia Madagascar Malawi Malaysia Maldives Marshall Islands Mauritius Mexico Micronesia Mongolia Mozambique Namibia Nauru Nepal New Zealand Niger Nigeria Oman Palau Palestine Panama Papua New Guinea Paraguay Peru Philippines Qatar Rwanda Saint Kitts and Nevis Saint Lucia Saint Vincent and the Grenadines Samoa São Tomé and Príncipe Senegal Seychelles Sierra Leone Singapore Solomon Islands South Africa South Korea Sri Lanka Suriname Taiwan Thailand Tonga Trinidad and Tobago Tuvalu Ukraine United Arab Emirates United Kingdom United States Uruguay Vanu…
Skip to main contentSkip to footer , Google Cloud (Vertex AI), and Microsoft Azure (Foundry). []( ## Related content ### Introducing Claude Design by Anthropic Labs Today, we’re launching Claude Design, a new Anthropic Labs product that lets you collaborate with Claude to create polished visual work like designs, prototypes, slides, one-pagers, and more. Read more ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. Read more ### Anthropic’…
Long context Claude Opus 4.7 has a 1M token context window. ## Optimizing model performance Please refer to the Claude Opus 4.7 system card . ## Additional assets Claude Documentation : Visit Claude documentation for a wealth of resources on model capabilities, prompting techniques, use case guidelines, and more. Adaptive Thinking Guide : Understand how best to use adaptive thinking with Claude. Claude Prompting Resources : Check out Anthropic's prompting tools and guides to learn how to craft prompts that elicit more helpful, nuanced responses. Claude Cookbooks : Check out example code for…
Pricing for Opus 4.7 starts at $5 per million input tokens and $25 per million output tokens, with up to 90% cost savings with prompt caching and 50% savings with batch processing. To learn more, check out our pricing page. To get started, use i.j4i.i2
claude-opus-4-7
via the Claude API. For workloads that need to run in the US, US-only inference is available at 1.1x pricing for input and output tokens. Learn more. ## Use cases Opus 4.7 is a premium model that works best for tasks no prior model could handle and where performance matters most. It’s built for professional software engineering, comple…
Anthropic: Claude Opus 4.7 ### anthropic/claude-opus-4.7 Released Apr 16, 20261,000,000 context$5/M input tokens$25/M output tokens Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on the coding and agentic strengths of Opus 4.6, it delivers stronger performance on complex, multi-step tasks and more reliable agentic execution across extended workflows. It is especially effective for asynchronous agent pipelines where tasks unfold over time - large codebases, multi-stage debugging, and end-to-end project orchestration. [...] Be…
Get started Get started Anthropic Anthropic # Claude Opus 4.7 Claude Opus 4.7 is a reasoning model by Anthropic with vision, function calling, prompt caching, structured output — designed for complex, multi-step problem solving where accuracy matters more than speed. It supports a 1M token context window with up to 128K tokens of output. Priced at $5 per million input tokens and $25 per million output tokens. Access Claude Opus 4.7 through Inworld Router with OpenAI SDK compatibility, built-in failover, and intelligent routing across providers. →Function CallingReasoningPrompt CachingResponse…
At a spec level, Opus 4.7 is positioned as Anthropic’s most capable generally available model for coding, enterprise workflows, multimodal reasoning, financial analysis, life sciences, cybersecurity, and long-running agentic work. It supports a 1M context window with no long-context pricing premium, up to 128K output tokens, and standard Opus pricing at $5 per million input tokens and $25 per million output tokens. The model's reliable knowledge cutoff is January 2026. [...] ## Same Price, Different Token Economics The first thing to understand about Opus 4.7 is that list pricing is unchanged…
logo # Claude Opus 4.7 Price: 2026 API Rates & Subscription avatar Claude Opus 4.7 Price: 2026 API Rates & Subscription Claude Opus 4.7 officially retails at $20/month for the Claude Pro plan, while API rates hold steady at $5.00 per million input tokens and $25.00 per million output tokens. However, despite these stable sticker prices, users face a critical technical hurdle: the new 2026 tokenizer produces up to 35% more tokens for identical raw text compared to version 4.6. This "tokenizer inflation," coupled with the high latency of Adaptive Thinking reasoning loops, means your effective p…
Claude Opus 4.7 Pricing: The Real Cost Story Behind the “Unchanged” Price Tag Press Option+1 for screen-reader mode, Option+0 to cancelAccessibility Screen-Reader Guide, Feedback, and Issue Reporting | New window | Output ($/1M) | Context | Best for | --- --- | Claude Opus 4.7 | $5 | $25 | 1M tokens | Frontier coding, agents, high-res vision | | Claude Opus 4.6 | $5 | $25 | 1M tokens | Still-capable coding, lower effective cost/request | | Claude Sonnet 4.6 | $3 | $15 | 1M tokens | Default for most production inference | | Claude Haiku 4.5 | $1 | $5 | 200K tokens | High-volume, low-latency,…
TL;DR Opus 4.7 是新旗舰。 Anthropic 已明确将其作为当前最强的通用可用 Opus 模型。 官方标价没变。 4.7 和 4.6 都是 i.j4i.i2
$5 / $25
。 真实任务成本不一定相同。 tokenizer 和 effort 行为变了，token 消耗可能上升。 迁移不是只改 model ID。 thinking、sampling、reasoning 展示方式都变了。 对编码和 agent 工作流来说，4.7 更值得优先测试。 ## 快速对比 [...] ### Claude Opus 4.7 的 model ID 是什么？ Claude API 的模型 ID 是 i.j4i.i2
claude-opus-4-7
。 ### Claude Opus 4.7 会比 Claude Opus 4.6 更贵吗？ 官方单价不变，都是 i.j4i.i2
$5 / MTok
输入和 i.j4i.i2
$25 / MTok
输出。但 Anthropic 明确表示，4.7 的 tokenizer 可能让同样输入消耗更多 token，所以实际任务成本仍可能上升。 ### Claude Opus 4.7 还支持 1M 上下文吗？ 支持。Anthropic 当前模型文档仍列出 i.j4i.i2
Claude Opus 4.7 发布：更费token了，用户评价两极 【TechWeb】4月17日消息，Anthropic宣布去最新模型 Claude Opus 4.7 已全面可用。相比 Opus 4.6，新模型在高级软件工程领域实现了显著进步，在多模态能力上视觉能力也大幅增强能产出更高质量的界面、幻灯片等。 Opus 4.7 的定价与 Opus 4.6 相同，每百万输入 token 5 美元，每百万输出 token 25 美元。 Anthropic在官方在公告里明确写道：新分词器会导致相同内容的token数变为原来的1.0到1.35倍。而API单价没变，实际成本就是变相上涨了10%-35%。 目前，已经有不少用户连夜试用了最新的Claude Opus 4.7，从社交平台上的信息来看，用户对Opus 4.7的评价较为两极。 大部分用户认可Opus 4.7在编程能力的提升，但是在文案撰写、对话沟通方面槽点较多。 Claude Opus 4.7主要能力提升 根据Anthropic官方贴出来的数据，在编码与推理方面，Opus 4.7 在处理复杂、长期运行的任务时更加严谨和一致，能够自行验证输出结果。用户反馈称，原本需要密切监督的复杂编码工作，现在可以放心交给 Opus 4.7。该模型能够严谨且一致地处理长时间、多步骤的任务，精确遵循指令，并在汇报前主动验证自身输出。 Opus 4.…
April 16, 2026 — Anthropic released Claude Opus 4.7 today, and the story is not about a new pricing tier or a dramatic architectural overhaul. It is about targeted, measurable improvements in the two areas that matter most for production use: coding and vision. The model scores 70% on CursorBench (up from 58%), achieves 98.5% visual-acuity (up from 54.5%), and solves 3x more production tasks than Opus 4.6 — all at the same $5/$25 per million token pricing. The API identifier is i.j4i.i2
claude-opus-4-7
. It is generally available across Claude products, the Anthropic API, Amazon Bedrock, Google Cloud…
AI Startup Shield AI Raises $1.5B in Series G Funding人工智能初创公司Shield AI完成15亿美元G轮融资La startup IA Shield AI lève 1,5 milliard de dollars lors d'un financement de série Gشركة الذكاء الاصطناعي Shield AI تجمع 1.5 مليار دولار في جولة تمويل من الفئة G #### Global Robotaxi Deployment Intensifies as U.S. and Chinese Firms Compete for Leadership全球无人驾驶出租车市场竞争加剧，美国和中国企业争夺主导地位Le Déploiement Mondial de Robotaxi S'intensifie Alors que les Entreprises Américaines et …تصاعد انتشار سيارات الأجرة الروبوتية عالميًا مع تنافس الشركات الأمريكية والصينية على القي… #### Sora Is Shutting Down: Transitioning Your A…
Additionally, the model "thinks more at higher effort levels," generating more output tokens during reasoning. At i.j4i.i2
xhigh
or i.j4i.i2
max
, Anthropic recommends setting i.j4i.i2
max_tokens
to at least 64K as a starting point. ## What's Unchanged Context window: 1M tokens, same as Opus 4.6, no long-context premium Output limit: 128K tokens per response Adaptive thinking and Agent Teams: both carry forward Prompt caching: up to 90% cost reduction on cached content Pricing structure: $5/$25 per million tokens across all platforms One notable removal: prefilling assistant messages now returns a 400 error on Opu…
Claude Opus 4.7深夜炸场，胜任更长任务、自主检查，视觉能力拉满 智东西4月17日报道，昨天夜间，Anthropic发布新一代旗舰大模型Claude Opus 4.7。 ▲Anthropic发布新模型Claude Opus 4.7（图源：X） 该模型在高级软件工程方面相比Opus 4.6有显著提升，尤其在处理最复杂的任务时提升明显；高分辨率图像处理能力大幅提升，是此前Claude模型的3倍以上；此外，Claude Code还同步新增了/ultrareview代码审查命令，输入后会启动审查会话，逐行检查代码变更。 用户反馈称，他们可以放心地将最难的编码工作交给Opus 4.7处理。Opus 4.7能够严谨一致地处理复杂的长时间运行任务，精确遵循指令，并在汇报结果之前自行验证输出。 Opus 4.7今日起在所有Claude产品和API、Amazon Bedrock、谷歌云Vertex AI以及Microsoft Foundry上线。定价与Opus 4.6一致：输入每百万token 5美元（约合人民币34元），输出每百万token 25美元（约合人民币170.5元）。开发者可通过Claude API使用claude-opus-4-7。 不得不说，Claude最近更新实在是快，大家都跟不上了，网友在Claude的评论区下面刷起了表情包，“两眼一睁，Claude又更新了”。…
Claude Opus 4.7深夜炸场，胜任更长任务、自主检查，视觉能力拉满 智东西4月17日报道，昨天夜间，Anthropic发布新一代旗舰大模型Claude Opus 4.7。 ▲Anthropic发布新模型Claude Opus 4.7（图源：X） 该模型在高级软件工程方面相比Opus 4.6有显著提升，尤其在处理最复杂的任务时提升明显；高分辨率图像处理能力大幅提升，是此前Claude模型的3倍以上；此外，Claude Code还同步新增了/ultrareview代码审查命令，输入后会启动审查会话，逐行检查代码变更。 用户反馈称，他们可以放心地将最难的编码工作交给Opus 4.7处理。Opus 4.7能够严谨一致地处理复杂的长时间运行任务，精确遵循指令，并在汇报结果之前自行验证输出。 Opus 4.7今日起在所有Claude产品和API、Amazon Bedrock、谷歌云Vertex AI以及Microsoft Foundry上线。定价与Opus 4.6一致：输入每百万token 5美元（约合人民币34元），输出每百万token 25美元（约合人民币170.5元）。开发者可通过Claude API使用claude-opus-4-7。 不得不说，Claude最近更新实在是快，大家都跟不上了，网友在Claude的评论区下面刷起了表情包，“两眼一睁，Claude又更新了”。…
然而，对于广大的中国大陆用户来说，如何顺畅、安全地使用 Claude，始终是一个核心痛点。本篇深度指南将为您一站式解决所有问题，内容涵盖： Claude 是什么及其核心优势。 国内访问 Claude 的两种主要路径。 Claude 核心模型（Haiku, Sonnet, Opus）全方位对比。 Claude Pro 会员的价值与订阅方法。 数据安全与合规使用的建议。 ### 1. Claude 是什么？为什么它如此特别？# Claude 是由前 OpenAI 核心成员创立的 Anthropic 公司开发的系列大型语言模型。相比其他模型，Claude 的设计哲学更侧重于 AI 安全与可预测性。其核心优势体现在： 超长的上下文窗口：Claude 4.5 提供高达 200K token 的上下文窗口，相当于可以一次性“阅读”并分析一本约 15 万单词的书。这使其在处理法律文件、财务报告、学术论文等长篇文档时具有无与伦比的优势。 顶级的推理与分析能力：在多项基准测试中，Claude 的旗舰模型（如 Opus 级）在研究生水平的推理、数学和编程方面均表现出顶级性能。 更少的“幻觉”与更高的诚实度：Claude 在被设计时就被要求尽可能避免编造信息。当它不知道答案时，更倾向于承认自己的局限性，而不是胡编乱造。 强大的多语言能力：Claude 对包括中文在内的多种语言都提供了出色的支持，能够理…
Sharadha Kandasubramanian Sharadha Kandasubramanianis a Senior Technical Program Manager for Amazon Bedrock. She drives cross-functional GenAI programs for Amazon Bedrock, enabling customers to grow and scale their GenAI workloads. Outside of work, she’s an avid runner and biker who loves spending time outdoors in the sun. Gareth ### Gareth Jones Gareth Jones is a Product Manager at Anthropic focused on the Claude API. His work includes collaboration with AWS on Amazon Bedrock to make Claude widely accessible to developers. Loading comments… ### Resources ### Blog Topics ### Follow ## Lea…
Claude 3 Supported and Unsupported Countries Unlike what was before, in 2025, Claude is now available in most parts of the world. This includes US, Canada, the United Kingdom, nearly all of Europe, Australia, Japan, South Korea, other countries across Asia, Africa, and Latin America. Anthropic has made it clear that Claude is not available in mainland China or a few other restricted places due to local laws and security rules. People and companies in supported countries can use Claude through its website, API, or tools like Claude Code (claude code might not be available in your country mis…
bash aws bedrock-runtime invoke-model \ --model-id us.anthropic.claude-opus-4-7 \ --region us-east-1 \ --body '{"anthropic_version":"bedrock-2023-05-31", "messages": [{"role": "user", "content": "Design a distributed architecture on AWS in Python that should support 100k requests per second across multiple geographic regions."}], "max_tokens": 32000}' \ --cli-binary-format raw-in-base64-out \ invoke-model-output.txt
Bash For more intelligent reasoning capability, you can use Adaptive thinking with Claude Opus 4.7, which lets Claude dynamically allocate thinking token budgets based on t…
For Enterprise organizations, Claude Design is off by default. Admins can enable it in Organization settings. Start designing at claude.ai/design. []( ## Related content ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. Read more ### Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors Read more ### Anthropic expands partnership with Google and Broadcom for multiple gigawatts of next-generation compute Read m…
Methodology All evals were run with a 64K thinking budget, interleaved scratchpads, 200K context window, default effort (high), default sampling settings (temperature, top_p), and averaged over 5 independent trials. Exceptions: SWE-bench Verified (no thinking budget) and Terminal Bench (128K thinking budget). Please see the Claude Opus 4.5 system card for full details. []( ## Related content ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters mo…
Context compaction (beta). Long-running conversations and agentic tasks often hit the context window. Context compaction automatically summarizes and replaces older context when the conversation approaches a configurable threshold, letting Claude perform longer tasks without hitting limits. 1M token context (beta). Opus 4.6 is our first Opus-class model with 1M token context. Premium pricing applies for prompts exceeding 200k tokens ($10/$37.50 per million input/output tokens), available only on the Claude Platform. 128k output tokens. Opus 4.6 supports outputs of up to 128k tokens, which let…
Image 21: logo > Claude Opus 4.7 is the most capable model we've tested at Quantium. Evaluated against leading AI models through our proprietary benchmarking solution, the biggest gains showed up where they matter most: reasoning depth, structured problem-framing, and complex technical work. Fewer corrections, faster iterations, and stronger outputs to solve the hardest problems our clients bring us. > > Ben Chan > > Chief AI Officer Image 22: logo [...] Image 13: logo > Claude Opus 4.7 is a very impressive coding model, particularly for its autonomy and more creative reasoning. On CursorBenc…
The changes mark an escalation from previous measures that targeted companies located within China. Dario Amodei, CEO and co-founder of Anthropic. Photo: AP alt=Dario Amodei, CEO and co-founder of Anthropic. Photo: AP> Anthropic, the developer of the Claude AI models known for their coding capabilities, is not the first US AI start-up to restrict access from mainland China and Hong Kong. Last July, OpenAI also tightened its policies to block access from "unsupported countries and territories". Meanwhile, China has implemented a registration system for AI services. Currently, all approved mode…
Related content ### Introducing Claude Design by Anthropic Labs Today, we’re launching Claude Design, a new Anthropic Labs product that lets you collaborate with Claude to create polished visual work like designs, prototypes, slides, one-pagers, and more. ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. ### Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors ### Products ### Models ### Solutions ### Clau…
| Jurisdiction | Claude.ai Access | Reason for Restriction | API Access | --- --- | | Hong Kong | Blocked | Anthropic regional access policy, NSL risk environment | Not supported | | Mainland China | Blocked | Anthropic policy plus domestic AI registration requirements | Not supported | | Taiwan | Available | Supported region | Supported | | Singapore | Available | Supported region | Supported | | Japan | Available | Supported region | Supported | ## How the Block Works Technically [...] Potentially, but with important caveats. Anthropic's September 2025 terms update blocked access for any co…
上班族為什麼愈來愈常選Claude做簡報？問題從這裡開始有趣起來。ChatGPT在文字問答上依然強大，但論到視覺設計感、排版邏輯、把一大段文字轉化成有結構美感的投影片，Claude用戶口碑明顯比前者高出一截。特別是Claude的Artifacts功能——讓模型直接在對話視窗裡生成可預覽的HTML或互動內容——讓許多設計力有限的上班族直接省掉了還要把生成內容轉去其他軟體的麻煩。 延伸閱讀：Claude是什麼？Artifacts怎麼用？ 根據Anthropic官網資訊，4月16日推出的Claude Opus 4.7，是Claude系列目前最高階的模型，Anthropic官方特別強調這一代在視覺美學品味上的提升，聲勢大到還沒正式上線，Adobe和Figma股價就先跌了一波。 Claude也支援繁體中文，包括台灣用語、正式商業文件、以及口語化的說明文字都能流暢處理。若需要確保輸出使用繁體，指令裡明確加上「請使用繁體中文」即可。 ## Claude Opus 4.7適合哪些人用？ Claude Opus 4.7的功能廣，但有幾種人特別會用到它的核心優勢： ### 常做提案簡報的上班族 業務、行銷、專案管理人員，每週需要產出提案、週報、成果報告。Opus 4.7新版Max output上限從64K tokens提升至128K tokens，其實就是一次能生成更長的內容而不被截斷。過去模型往往…
Read more ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. Read more ### Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors Read more []( ### Products Claude Claude Code Claude Code Enterprise Claude Code Security Claude Cowork Claude for Chrome Claude for Slack Claude for Excel Claude for PowerPoint Claude for Word Skills Max plan Team plan Enterprise plan Download app Pricing Log in to Claude ### Models…
3.10 Adaptive thinking only Fixed thinking budgets are gone. Use i.j4i.i2
thinking: {type: 'adaptive'}
plus the i.j4i.i2
effort
parameter. Old API calls with i.j4i.i2
budget_tokens
return HTTP 400. Find and replace before flipping the model flag: i.j4i.i2
client = anthropic.Anthropic() response = client.messages.create( model="claude-opus-4-7", max_tokens=16000, thinking={"type": "adaptive"}, messages=[ { "role": "user", "content": "Explain why the sum of two even numbers is always even.", } ], )
If you’re not an engineer, you don’t need to memorize the syntax. Just give your agent this documentation: (remove…
Share US AI giant Anthropic bars Chinese-owned entities The startup, heavily backed by Amazon, is known for its Claude chatbot and positions itself as focused on AI safety and responsible development. Companies based in China, as well as in countries including Russia, North Korea and Iran, are already unable to access Anthropic's commercial services over legal and security concerns. ChatGPT and other products from US competitor OpenAI are also unavailable within China -- spurring the growth of homegrown AI models from Chinese companies such as Alibaba and Baidu. Anthropic said in a statement…
“This update prohibits companies or organisations whose ownership structures subject them to control from jurisdictions where our products are not permitted, like China, regardless of where they operate,” it said. The changes mark an escalation from previous measures that targeted companies located within China. Dario Amodei, CEO and co-founder of Anthropic. Photo: AP Meanwhile, China has implemented a registration system for AI services. Currently, all approved models are domestically developed. The country has a vibrant open-source community for AI models, including those from DeepSeek and…
web browsers, and user-focused workflows like email management. We implemented several protective measures to combat prompt injection attacks, including specialized reinforcement learning training to help the model recognize and avoid these manipulations, and the deployment of detection systems that can halt the model’s execution when a potential injection attempt is identified. These end-to-end defenses improved both models' prompt injection safety scores compared to their performance without safeguards (see table 3.2A). Model Attack prevention score (without safeguards) Attack prevention sc…
Dario Amodei, CEO and co-founder of Anthropic. Photo via AP. ### AI and Autonomy # Anthropic blocks Chinese-owned firms from AI services over security concerns This move marks the first instance of a major U.S.-based AI company denying access to its technology over national security concerns By Joseph Morton Published Anthropic, the San Francisco-based developer of the Claude AI model, has announced it will cease providing artificial intelligence services to organizations majority-owned by Chinese entities. This policy, effective immediately, also extends to subsidiaries operating in other co…
logo AI编程社区 ## 登录社区云 ### AI编程社区 欢迎加入社区 欢迎加入社区 ## 温馨提示：您尚未绑定手机号 为遵守国家网络实名制规定，未绑定将限制内容发布与互动 立即绑定 AI编程社区 Claude AI国内使用完全指南：免费体验Claude 4最强AI编程助手 # Claude AI国内使用完全指南：免费体验Claude 4最强AI编程助手 Claude AI是由Anthropic公司开发的大型语言模型，具有安全、可靠等技术优势。最新版本Claude 4在代码理解、数学推理等方面有显著提升。相比ChatGPT，Claude在上下文长度、中文理解等方面表现更优。国内用户可通过aicodewith.com平台稳定访问Claude AI，享受官方授权接口、中文界面等便利。平台提供Claude 4、Sonnet等全系列模型，支持编程辅助、内容 训练方法，在安全性、可靠性和有用性方面表现突出。 Claude 产品家族包括： Claude Sonnet：平衡性能与成本的主流版本 Claude Opus：最高性能的旗舰版本 Claude Code：专为编程任务优化的版本 Claude Desktop：桌面应用程序 #### 1.2 Claude 4：技术革新的里程碑 最新发布的 claude 4 在多个维度实现了突破性提升： | 能力维度 | Claude 4 | 前代…
Models Opus Sonnet Haiku Log in Claude.ai Claude Console EN This is some text inside of a div block. Log in to ClaudeLog in to Claude Log in to Claude Download appDownload app Download app # AI research and products that put safety at the frontier AIresearchandproductsthat put safety at the frontier AI will have a vast impact on the world. Anthropic is a public benefit corporation dedicated to securing its benefits and mitigating its risks. ## Project Glasswing Securing critical software for the AI era Continue reading Image 1 Read the storyRead the story ## Latest releases ### Claude Opus 4.…
Skip to main contentSkip to footer | October 2024 | Read system card | | Claude Sonnet 3.5 | June 2024 | Read system card | | Claude 3 | March 2024 | Read system card | | Claude 2 | July 2023 | Read system card | []( ### Products Claude Claude Code Claude Code Enterprise Claude Code Security Claude Cowork Claude for Chrome Claude for Slack Claude for Excel Claude for PowerPoint Claude for Word Skills Max plan Team plan Enterprise plan Download app Pricing Log in to Claude ### Models Mythos preview Opus Sonnet Haiku ### Solutions AI agents Code modernization Coding Customer support Education F…
Apr 17, 2026 Product Introducing Claude Design by Anthropic Labs Apr 16, 2026 Product Introducing Claude Opus 4.7 Apr 14, 2026 Announcements Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors Apr 6, 2026 Announcements Anthropic expands partnership with Google and Broadcom for multiple gigawatts of next-generation compute Mar 31, 2026 Announcements Australian government and Anthropic sign MOU for AI safety and research Mar 12, 2026 Announcements Anthropic invests $100 million into the Claude Partner Network Mar 11, 2026 Announcements Introducing The Anthropic Ins…
Afghanistan Belarus Central African Republic China Congo, Democratic Republic of Cuba Eritrea Ethiopia Iran Libya Mali Myanmar Nicaragua North Korea Russia Somalia Sudan Syria Venezuela Yemen
I took their list here and diff'ed it with Wikipedia's list of sovereign states. They don't accept the Democratic Republic of Congo, but the Republic of Congo is OK. There are indeed two Congos with ridiculously similar names, and apparently the one that calls itself democratic is the bad one. Anthropic is fine with Taiwan. (They give two lists, one for API and another for Claude.ai, but the list…
Model descriptionClaude Opus 4 and Claude Sonnet 4 are two new hybrid reasoning large language models from Anthropic. They have advanced capabilities in reasoning, visual analysis, computer use, and tool use. They are particularly adept at complex computer coding tasks, which they can productively perform autonomously for sustained periods of time. In general, the capabilities of Claude Opus 4 are stronger than those of Claude Sonnet 4. Benchmarked CapabilitiesSee our Claude 4 Announcement Acceptable UsesSee our Usage Policy Release dateMay 2025 Access SurfacesClaude Opus 4 and Claude Sonnet…
Models Opus Sonnet Haiku ### Solutions AI agents Code modernization Coding Customer support Education Financial services Government Healthcare Life sciences Nonprofits Security ### Claude Platform Overview Developer docs Pricing Marketplace Amazon Bedrock Google Cloud’s Vertex AI Microsoft Foundry Regional compliance Console login ### Resources Blog Claude partner network Connectors Courses Customer stories Engineering at Anthropic Events Plugins Powered by Claude Service partners Community Campus Program Startups program Tutorials Use cases ### Company [...] ### Models Opus Sonnet Haiku…
| State | Type of Restrictions | Duration | Number of Announced Data Centers | --- --- | | Georgia | Temporary ban | Until March 2027 | 340 | | Maryland | Conditional restrictions | Not fixed | 10 | | Michigan | Temporary ban | Not specified | 21 | | New Hampshire | Temporary ban | 1 year | 0 | | New York | Temporary ban | 3 years | 72 | | Oklahoma | Temporary ban | Until Nov 2029 | 34 | | South Carolina | Temporary ban | Until Jan 2028 | 8 | | South Dakota | Temporary ban | 1 year | 6 | | Vermont | Temporary ban | Until July 2030 | 0 | | Virginia | Conditional restrictions | Not fixed | 498…
Mauritius Mexico Micronesia Moldova Monaco Mongolia Montenegro Morocco Mozambique Namibia Nauru Nepal Netherlands New Zealand Niger Nigeria North Macedonia Norway Oman Pakistan Palau Palestine Panama Papua New Guinea Paraguay Peru Philippines Poland Portugal Qatar Romania Rwanda Saint Kitts and Nevis Saint Lucia Saint Vincent and the Grenadines Samoa San Marino São Tomé and Príncipe Saudi Arabia Senegal Serbia Seychelles Sierra Leone Singapore Slovakia Slovenia Solomon Islands South Africa South Korea Spain Sri Lanka Suriname Sweden Switzerland Taiwan Tajikistan Tanzania Thailand Timor-Leste…
答案公共网址2026年4月20日100 来源
Show me top 10 trending question Chinese Mainland users often ask about Claude Opus 4.7 now. Show me both Chinese Mainland (zh CN) language & English version for me to understand these questions.
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent perform...
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and latency, and the workflow changes needed to use it well. I can’t verify exact Main
要点
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and la
I can’t verify exact Mainland search volume rankings from open sources, so this is a best effort trend synthesis rather than an official leaderboard.
I’ve written each question in simplified Chinese and English so you can scan them quickly.
[13][16] Key findings 1.
zh CN：大陆现在能直接注册并稳定使用 Claude Opus 4.7 吗？ English: Can users in Mainland China register for and use Claude Opus 4.7 reliably right now?
zh CN：如果我现在用的是 Opus 4.6，升级到 4.7 到底值不值？ English: If I’m using Opus 4.6 now, is upgrading to 4.7 actually worth it?
研究答案
Based on currently indexed Chinese-language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and latency, and the workflow changes needed to use it well. I can’t verify exact Mainland search-volume rankings from open sources, so this is a best-effort trend synthesis rather than an official leaderboard. I’ve written each question in simplified Chinese and English so you can scan them quickly. [13][16]
Key findings
zh-CN：大陆现在能直接注册并稳定使用 Claude Opus 4.7 吗？
English: Can users in Mainland China register for and use Claude Opus 4.7 reliably right now? [8][11]
zh-CN：如果我现在用的是 Opus 4.6，升级到 4.7 到底值不值？
English: If I’m using Opus 4.6 now, is upgrading to 4.7 actually worth it? [12][13][16]
zh-CN：4.7 在编程、前端开发和长任务 Agent 上到底强多少？
English: How much better is 4.7 for coding, frontend work, and long-running agent tasks? [12][16]
zh-CN：为什么很多人说 4.7 更会“接情绪”了，但写作风格反而变差了？
English: Why do many users say 4.7 feels more emotionally accommodating, but worse in writing style? [16]
zh-CN：官方价格看起来没明显涨，为什么实际感觉更费 tokens、更慢、更贵？
English: If the official pricing doesn’t look much higher, why does it feel more token-hungry, slower, and costlier in practice? [10][13][16]
zh-CN：在 Claude Code 或 API 里，adaptive thinking、effort、xhigh 应该怎么设置？
English: In Claude Code or the API, how should adaptive thinking, effort, and xhigh be configured? [7][13]
zh-CN：1M 上下文窗口到底有没有实际价值，最适合哪些场景？
English: How useful is the 1M-context window in real work, and which use cases benefit most? [9][12]
zh-CN：4.7 的视觉能力到底提升了多少，能不能更稳地看 UI、截图、图表和 PDF？
English: How much did 4.7’s vision improve, and is it more reliable on UIs, screenshots, charts, and PDFs? [12][16]
zh-CN：从 4.6 迁移到 4.7，会踩到哪些 API、参数或提示词层面的坑？
English: What API, parameter, or prompting pitfalls appear when migrating from 4.6 to 4.7? [13]
zh-CN：4.7 的安全限制是不是更严了，会不会更容易拒答或拦截高风险请求？
English: Are 4.7’s safety restrictions stricter, and will it refuse or block high-risk requests more often? [13]
Confirmed facts
Official Claude docs say users should check supported regions for API availability, so location and access are a real practical concern behind many user questions. [8][11]
Official docs say Claude Opus 4.7 has a 1M-token context window and present it as improved on knowledge work, vision, and long-context tasks. [9][12]
Official migration docs say Opus 4.7 shifts to adaptive thinking, changes tokenization, removes some prior parameter behavior, and tends to use tools less by default. [7][13]
Official pricing docs list Opus 4.7 pricing alongside Opus 4.6, while Chinese launch coverage says the pricing strategy stayed the same as the prior version. [10][16]
Chinese-language coverage around launch highlights stronger coding and long-task reliability, but also surfaces complaints about writing tone, search feel, and overall “personality.” [16]
What remains inference
This is not a true search-volume ranking from Baidu, WeChat, Xiaohongshu, or internal Anthropic analytics.
The ordering is inferred from repeated themes in indexed Chinese-language coverage and developer discussion, which likely overrepresents technical users. [13][16]
What the evidence suggests
The biggest attention split is between access questions and “is 4.7 actually better for my workflow than 4.6?” questions. [8][11][13][16]
Developer-heavy questions dominate because the official changes are workflow-heavy: adaptive thinking, effort tuning, long context, migration changes, and tool-use behavior. [7][9][12][13]
The curiosity is not just about benchmark gains. People also care about whether daily use feels worse in style, speed, or cost. [13][16]
Conflicting evidence or uncertainty
Official docs frame 4.7 as stronger on coding, knowledge work, vision, and long-context use. [12]
Chinese-language commentary still reports mixed subjective reactions, especially around writing tone, search quality, and whether the model feels less distinctive. [16]
So the safest conclusion is that 4.7 looks especially compelling for coding and agentic workflows, but not everyone sees it as a universal upgrade for writing-heavy use. [12][16]
Open questions
How much of the negative reaction comes from the base model versus app-layer behavior, safety layers, or prompting habits? [13][16]
How are Mainland users actually accessing it in practice: direct account use, partner clouds, or other workflows? [8][11]
Is the perceived writing-style regression fixable with prompting and effort tuning, or is it a real tradeoff in this release? [7][13][16]
Sources worth trusting most
Official Claude docs on what changed in Opus 4.7, migration behavior, context windows, pricing, and supported regions. [7][9][10][11][12][13]
Chinese-language launch and reaction coverage for what users are actively debating right now. [16]
Recommended next step
If you want, I can next turn these 10 questions into a bilingual FAQ with short, factual answers in zh-CN + English, so you can use it for research, content planning, customer support, or market monitoring.
Summary
Your best current “top 10” for Mainland-facing Chinese discussion is: access, upgrade value, coding gains, writing-style complaints, token cost, adaptive-thinking setup, 1M context usefulness, vision gains, migration pitfalls, and stricter safety behavior. That pattern is well supported by the combination of official docs and current Chinese-language discussion, but the exact rank order is still an inference rather than hard search-volume proof. [8][9][12][13][16]
人们还问
“Show me top 10 trending question Chinese Mainland users often ask about Claude Opus 4.7 now. Show me both Chinese Mainland (zh CN) language & English version for me to understand these questions.”的简短答案是什么？
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and la
首先要验证的关键点是什么？
Based on currently indexed Chinese language coverage and developer discussion, the hottest Claude Opus 4.7 questions are mostly about Mainland access, whether it is truly better than Opus 4.6, coding and agent performance, token cost and la I can’t verify exact Mainland search volume rankings from open sources, so this is a best effort trend synthesis rather than an official leaderboard.
接下来在实践中我应该做什么？
I’ve written each question in simplified Chinese and English so you can scan them quickly.
Structure document content and metadata with XML tags: When using multiple documents, wrap each document in i.j4i.i2
<document>
tags with i.j4i.i2
<document_content>
and i.j4i.i2
<source>
(and other metadata) subtags for clarity. i.j4i.i2
<document>
i.j4i.i2
<document_content>
i.j4i.i2
<source>
### Example multi-document structure Ground responses in quotes: For long document tasks, ask Claude to quote relevant parts of the documents first before carrying out its task. This helps Claude cut through the noise of the rest of the document's contents. ### Example quote extraction ### Model self-knowledge If you would like Claude to ident…
跳至主要內容 Google Cloud Documentation 控制台 English Deutsch Español Español – América Latina Français Indonesia Italiano Português Português – Brasil עברית 中文 – 简体 中文 – 繁體 日本語 한국어 登入 。 ,,["上次更新時間：2026-04-18 (世界標準時間)。"],[],[]]
Claude Opus 4.7 Text Generation • Anthropic • Proxied Claude Opus 4.7 is Anthropic's most capable generally available model to date. It is highly autonomous and performs exceptionally well on long-horizon agentic work, knowledge work, vision tasks, and memory tasks. | Model Info | | Context Window ↗ | 1,000,000 tokens | | Terms and License | link ↗ | | More information | link ↗ | ## Usage TypeScript ``` const response = await env.AI.run( 'anthropic/claude-opus-4.7', { messages: [ { role: 'user', content: 'What are the three laws of thermodynamics?', }, ], max_tokens: 1024, }, { gateway: { i…
February 12, 2026 Self-serve Enterprise plans Previously, Enterprise plans were only available to customers working with our Sales team. Now, any organization can purchase an Enterprise plan directly on our website with no Sales conversation required. Self-serve Enterprise plans have a single seat type that includes access to Claude, Claude Code, and Cowork. For more information, refer to our blog post or What is the Enterprise plan? ### February 5, 2026 Claude Opus 4.6 launch We’ve upgraded our smartest model and improved its coding skills. Read our blog post for more information: Introd…
Adaptive thinking is the recommended way to use extended thinking with Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6, and is the default mode on Claude Mythos Preview (where it auto-applies whenever i.j4i.i2
thinking
is unset). Instead of manually setting a thinking token budget, adaptive thinking lets Claude dynamically determine when and how much to use extended thinking based on the complexity of each request. On Claude Opus 4.7, adaptive thinking is the only supported thinking mode; manual i.j4i.i2
For detailed information about limits, tiers, and the token bucket algorithm used for rate limiting, see Rate limits. ### Availability The Claude API is available in many countries and regions worldwide. Check the supported regions page to confirm availability in your location. ## Next Steps Messages API reference Complete API specification for direct model interactionsClaude Managed Agents reference Agents, Sessions, and Environments endpointsClient SDKs Python, TypeScript, Java, Go, C#, Ruby, and PHPRate limits Usage tiers, spend limits, and token bucket algorithm Was this page helpful? [..…
Claude Mythos Preview, Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6 have a 1M-token context window. Other Claude models, including Claude Sonnet 4.5 and Sonnet 4 (deprecated), have a 200k-token context window. A single request can include up to 600 images or PDF pages (100 for models with a 200k-token context window). When sending many images or large documents, you may approach request size limits before the token limit. ## Context awareness in Claude Sonnet 4.6, Sonnet 4.5, and Haiku 4.5 [...] How it works: At the start of a conversation, Claude receives information about its tot…
Seychelles Sierra Leone Singapore Slovakia Slovenia Solomon Islands South Africa South Korea Spain Sri Lanka Suriname Sweden Switzerland Taiwan Tajikistan Tanzania Thailand Timor-Leste, Democratic Republic of Togo Tonga Trinidad and Tobago Tunisia Turkey Turkmenistan Tuvalu Uganda Ukraine (except Crimea, Donetsk, and Luhansk regions) United Arab Emirates United Kingdom United States of America Uruguay Uzbekistan Vanuatu Vietnam Zambia Zimbabwe [...] India Indonesia Iraq Ireland Israel Italy Jamaica Japan Jordan Kazakhstan Kenya Kiribati Kuwait Kyrgyzstan Laos Latvia Lebanon Lesotho Liberia Li…
parameters to give additional headroom, including compaction triggers. Claude Opus 4.7 provides a 1M context window at standard API pricing with no long-context premium. ## Capability improvements ### Knowledge work Claude Opus 4.7 shows meaningful gains on knowledge-worker tasks, particularly where the model needs to visually verify its own outputs: .docx redlining and .pptx editing — improved at producing and self-checking tracked changes and slide layouts. Charts and figure analysis — improved at programmatic tool-calling with image-processing librarie…
實時網路安全防護： Claude Opus 4.7 新增的功能，涉及禁止或高風險主題的請求可能導致拒絕。對於合法的安全工作，例如滲透測試、漏洞研究或紅隊，請申請 網路驗證計畫 以請求降低對網路內容的限制。有關背景，請參閱 防護、警告和上訴。 高解析度影像支援： Claude Opus 4.7 是第一個具有高解析度影像支援的 Claude 模型，長邊的最大影像解析度為 2576 像素（從先前模型的 1568 像素提高）。這為視覺密集型工作負載帶來了收益，對於電腦使用、螢幕截圖理解和文件分析特別有價值。高解析度支援是自動的，不需要測試版標頭或用戶端選擇加入。全解析度影像可以使用比先前模型多約 3 倍的影像代幣（每個影像最多 4,784 代幣，相比之前的大約 1,6…
We implemented several protective measures to combat prompt injection attacks, including specialized reinforcement learning training to help the model recognize and avoid these manipulations, and the deployment of detection systems that can halt the model’s execution when a potential injection attempt is identi fi ed. These end-to-end defenses improved both models' prompt injection safety scores compared to their performance without safeguards. (see table 3.2A). Model Attack prevention score (without safeguards) Attack prevention score (with safeguards) Claude Opus 4 71% 89% Claude Sonnet 4 6…
AsiaBAN AsiaBAN # Anthropic bans Chinese-controlled entities from accessing Claude models Credit: AFP US AI startup Anthropic has enforced a new policy prohibiting Chinese-controlled entities from using its Claude series models, prompting ByteDance's Singapore-based IDE platform Trae to immediately remove Claude from its offerings. The move has attracted... Keep me signed in Some subscribers prefer to save their log-in information so they do not have to enter their User ID and Password each time they visit the site. To activate this function, check the 'Keep me signed in' box in the log-in se…
Japan Kenya Kiribati Kuwait Lebanon Lesotho Liberia Madagascar Malawi Malaysia Maldives Marshall Islands Mauritius Mexico Micronesia Mongolia Mozambique Namibia Nauru Nepal New Zealand Niger Nigeria Oman Palau Palestine Panama Papua New Guinea Paraguay Peru Philippines Qatar Rwanda Saint Kitts and Nevis Saint Lucia Saint Vincent and the Grenadines Samoa São Tomé and Príncipe Senegal Seychelles Sierra Leone Singapore Solomon Islands South Africa South Korea Sri Lanka Suriname Taiwan Thailand Tonga Trinidad and Tobago Tuvalu Ukraine United Arab Emirates United Kingdom United States Uruguay Vanu…
Skip to main contentSkip to footer , Google Cloud (Vertex AI), and Microsoft Azure (Foundry). []( ## Related content ### Introducing Claude Design by Anthropic Labs Today, we’re launching Claude Design, a new Anthropic Labs product that lets you collaborate with Claude to create polished visual work like designs, prototypes, slides, one-pagers, and more. Read more ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. Read more ### Anthropic’…
Long context Claude Opus 4.7 has a 1M token context window. ## Optimizing model performance Please refer to the Claude Opus 4.7 system card . ## Additional assets Claude Documentation : Visit Claude documentation for a wealth of resources on model capabilities, prompting techniques, use case guidelines, and more. Adaptive Thinking Guide : Understand how best to use adaptive thinking with Claude. Claude Prompting Resources : Check out Anthropic's prompting tools and guides to learn how to craft prompts that elicit more helpful, nuanced responses. Claude Cookbooks : Check out example code for…
Pricing for Opus 4.7 starts at $5 per million input tokens and $25 per million output tokens, with up to 90% cost savings with prompt caching and 50% savings with batch processing. To learn more, check out our pricing page. To get started, use i.j4i.i2
claude-opus-4-7
via the Claude API. For workloads that need to run in the US, US-only inference is available at 1.1x pricing for input and output tokens. Learn more. ## Use cases Opus 4.7 is a premium model that works best for tasks no prior model could handle and where performance matters most. It’s built for professional software engineering, comple…
Anthropic: Claude Opus 4.7 ### anthropic/claude-opus-4.7 Released Apr 16, 20261,000,000 context$5/M input tokens$25/M output tokens Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on the coding and agentic strengths of Opus 4.6, it delivers stronger performance on complex, multi-step tasks and more reliable agentic execution across extended workflows. It is especially effective for asynchronous agent pipelines where tasks unfold over time - large codebases, multi-stage debugging, and end-to-end project orchestration. [...] Be…
Get started Get started Anthropic Anthropic # Claude Opus 4.7 Claude Opus 4.7 is a reasoning model by Anthropic with vision, function calling, prompt caching, structured output — designed for complex, multi-step problem solving where accuracy matters more than speed. It supports a 1M token context window with up to 128K tokens of output. Priced at $5 per million input tokens and $25 per million output tokens. Access Claude Opus 4.7 through Inworld Router with OpenAI SDK compatibility, built-in failover, and intelligent routing across providers. →Function CallingReasoningPrompt CachingResponse…
At a spec level, Opus 4.7 is positioned as Anthropic’s most capable generally available model for coding, enterprise workflows, multimodal reasoning, financial analysis, life sciences, cybersecurity, and long-running agentic work. It supports a 1M context window with no long-context pricing premium, up to 128K output tokens, and standard Opus pricing at $5 per million input tokens and $25 per million output tokens. The model's reliable knowledge cutoff is January 2026. [...] ## Same Price, Different Token Economics The first thing to understand about Opus 4.7 is that list pricing is unchanged…
logo # Claude Opus 4.7 Price: 2026 API Rates & Subscription avatar Claude Opus 4.7 Price: 2026 API Rates & Subscription Claude Opus 4.7 officially retails at $20/month for the Claude Pro plan, while API rates hold steady at $5.00 per million input tokens and $25.00 per million output tokens. However, despite these stable sticker prices, users face a critical technical hurdle: the new 2026 tokenizer produces up to 35% more tokens for identical raw text compared to version 4.6. This "tokenizer inflation," coupled with the high latency of Adaptive Thinking reasoning loops, means your effective p…
Claude Opus 4.7 Pricing: The Real Cost Story Behind the “Unchanged” Price Tag Press Option+1 for screen-reader mode, Option+0 to cancelAccessibility Screen-Reader Guide, Feedback, and Issue Reporting | New window | Output ($/1M) | Context | Best for | --- --- | Claude Opus 4.7 | $5 | $25 | 1M tokens | Frontier coding, agents, high-res vision | | Claude Opus 4.6 | $5 | $25 | 1M tokens | Still-capable coding, lower effective cost/request | | Claude Sonnet 4.6 | $3 | $15 | 1M tokens | Default for most production inference | | Claude Haiku 4.5 | $1 | $5 | 200K tokens | High-volume, low-latency,…
TL;DR Opus 4.7 是新旗舰。 Anthropic 已明确将其作为当前最强的通用可用 Opus 模型。 官方标价没变。 4.7 和 4.6 都是 i.j4i.i2
$5 / $25
。 真实任务成本不一定相同。 tokenizer 和 effort 行为变了，token 消耗可能上升。 迁移不是只改 model ID。 thinking、sampling、reasoning 展示方式都变了。 对编码和 agent 工作流来说，4.7 更值得优先测试。 ## 快速对比 [...] ### Claude Opus 4.7 的 model ID 是什么？ Claude API 的模型 ID 是 i.j4i.i2
claude-opus-4-7
。 ### Claude Opus 4.7 会比 Claude Opus 4.6 更贵吗？ 官方单价不变，都是 i.j4i.i2
$5 / MTok
输入和 i.j4i.i2
$25 / MTok
输出。但 Anthropic 明确表示，4.7 的 tokenizer 可能让同样输入消耗更多 token，所以实际任务成本仍可能上升。 ### Claude Opus 4.7 还支持 1M 上下文吗？ 支持。Anthropic 当前模型文档仍列出 i.j4i.i2
Claude Opus 4.7 发布：更费token了，用户评价两极 【TechWeb】4月17日消息，Anthropic宣布去最新模型 Claude Opus 4.7 已全面可用。相比 Opus 4.6，新模型在高级软件工程领域实现了显著进步，在多模态能力上视觉能力也大幅增强能产出更高质量的界面、幻灯片等。 Opus 4.7 的定价与 Opus 4.6 相同，每百万输入 token 5 美元，每百万输出 token 25 美元。 Anthropic在官方在公告里明确写道：新分词器会导致相同内容的token数变为原来的1.0到1.35倍。而API单价没变，实际成本就是变相上涨了10%-35%。 目前，已经有不少用户连夜试用了最新的Claude Opus 4.7，从社交平台上的信息来看，用户对Opus 4.7的评价较为两极。 大部分用户认可Opus 4.7在编程能力的提升，但是在文案撰写、对话沟通方面槽点较多。 Claude Opus 4.7主要能力提升 根据Anthropic官方贴出来的数据，在编码与推理方面，Opus 4.7 在处理复杂、长期运行的任务时更加严谨和一致，能够自行验证输出结果。用户反馈称，原本需要密切监督的复杂编码工作，现在可以放心交给 Opus 4.7。该模型能够严谨且一致地处理长时间、多步骤的任务，精确遵循指令，并在汇报前主动验证自身输出。 Opus 4.…
April 16, 2026 — Anthropic released Claude Opus 4.7 today, and the story is not about a new pricing tier or a dramatic architectural overhaul. It is about targeted, measurable improvements in the two areas that matter most for production use: coding and vision. The model scores 70% on CursorBench (up from 58%), achieves 98.5% visual-acuity (up from 54.5%), and solves 3x more production tasks than Opus 4.6 — all at the same $5/$25 per million token pricing. The API identifier is i.j4i.i2
claude-opus-4-7
. It is generally available across Claude products, the Anthropic API, Amazon Bedrock, Google Cloud…
AI Startup Shield AI Raises $1.5B in Series G Funding人工智能初创公司Shield AI完成15亿美元G轮融资La startup IA Shield AI lève 1,5 milliard de dollars lors d'un financement de série Gشركة الذكاء الاصطناعي Shield AI تجمع 1.5 مليار دولار في جولة تمويل من الفئة G #### Global Robotaxi Deployment Intensifies as U.S. and Chinese Firms Compete for Leadership全球无人驾驶出租车市场竞争加剧，美国和中国企业争夺主导地位Le Déploiement Mondial de Robotaxi S'intensifie Alors que les Entreprises Américaines et …تصاعد انتشار سيارات الأجرة الروبوتية عالميًا مع تنافس الشركات الأمريكية والصينية على القي… #### Sora Is Shutting Down: Transitioning Your A…
Additionally, the model "thinks more at higher effort levels," generating more output tokens during reasoning. At i.j4i.i2
xhigh
or i.j4i.i2
max
, Anthropic recommends setting i.j4i.i2
max_tokens
to at least 64K as a starting point. ## What's Unchanged Context window: 1M tokens, same as Opus 4.6, no long-context premium Output limit: 128K tokens per response Adaptive thinking and Agent Teams: both carry forward Prompt caching: up to 90% cost reduction on cached content Pricing structure: $5/$25 per million tokens across all platforms One notable removal: prefilling assistant messages now returns a 400 error on Opu…
Claude Opus 4.7深夜炸场，胜任更长任务、自主检查，视觉能力拉满 智东西4月17日报道，昨天夜间，Anthropic发布新一代旗舰大模型Claude Opus 4.7。 ▲Anthropic发布新模型Claude Opus 4.7（图源：X） 该模型在高级软件工程方面相比Opus 4.6有显著提升，尤其在处理最复杂的任务时提升明显；高分辨率图像处理能力大幅提升，是此前Claude模型的3倍以上；此外，Claude Code还同步新增了/ultrareview代码审查命令，输入后会启动审查会话，逐行检查代码变更。 用户反馈称，他们可以放心地将最难的编码工作交给Opus 4.7处理。Opus 4.7能够严谨一致地处理复杂的长时间运行任务，精确遵循指令，并在汇报结果之前自行验证输出。 Opus 4.7今日起在所有Claude产品和API、Amazon Bedrock、谷歌云Vertex AI以及Microsoft Foundry上线。定价与Opus 4.6一致：输入每百万token 5美元（约合人民币34元），输出每百万token 25美元（约合人民币170.5元）。开发者可通过Claude API使用claude-opus-4-7。 不得不说，Claude最近更新实在是快，大家都跟不上了，网友在Claude的评论区下面刷起了表情包，“两眼一睁，Claude又更新了”。…
Claude Opus 4.7深夜炸场，胜任更长任务、自主检查，视觉能力拉满 智东西4月17日报道，昨天夜间，Anthropic发布新一代旗舰大模型Claude Opus 4.7。 ▲Anthropic发布新模型Claude Opus 4.7（图源：X） 该模型在高级软件工程方面相比Opus 4.6有显著提升，尤其在处理最复杂的任务时提升明显；高分辨率图像处理能力大幅提升，是此前Claude模型的3倍以上；此外，Claude Code还同步新增了/ultrareview代码审查命令，输入后会启动审查会话，逐行检查代码变更。 用户反馈称，他们可以放心地将最难的编码工作交给Opus 4.7处理。Opus 4.7能够严谨一致地处理复杂的长时间运行任务，精确遵循指令，并在汇报结果之前自行验证输出。 Opus 4.7今日起在所有Claude产品和API、Amazon Bedrock、谷歌云Vertex AI以及Microsoft Foundry上线。定价与Opus 4.6一致：输入每百万token 5美元（约合人民币34元），输出每百万token 25美元（约合人民币170.5元）。开发者可通过Claude API使用claude-opus-4-7。 不得不说，Claude最近更新实在是快，大家都跟不上了，网友在Claude的评论区下面刷起了表情包，“两眼一睁，Claude又更新了”。…
然而，对于广大的中国大陆用户来说，如何顺畅、安全地使用 Claude，始终是一个核心痛点。本篇深度指南将为您一站式解决所有问题，内容涵盖： Claude 是什么及其核心优势。 国内访问 Claude 的两种主要路径。 Claude 核心模型（Haiku, Sonnet, Opus）全方位对比。 Claude Pro 会员的价值与订阅方法。 数据安全与合规使用的建议。 ### 1. Claude 是什么？为什么它如此特别？# Claude 是由前 OpenAI 核心成员创立的 Anthropic 公司开发的系列大型语言模型。相比其他模型，Claude 的设计哲学更侧重于 AI 安全与可预测性。其核心优势体现在： 超长的上下文窗口：Claude 4.5 提供高达 200K token 的上下文窗口，相当于可以一次性“阅读”并分析一本约 15 万单词的书。这使其在处理法律文件、财务报告、学术论文等长篇文档时具有无与伦比的优势。 顶级的推理与分析能力：在多项基准测试中，Claude 的旗舰模型（如 Opus 级）在研究生水平的推理、数学和编程方面均表现出顶级性能。 更少的“幻觉”与更高的诚实度：Claude 在被设计时就被要求尽可能避免编造信息。当它不知道答案时，更倾向于承认自己的局限性，而不是胡编乱造。 强大的多语言能力：Claude 对包括中文在内的多种语言都提供了出色的支持，能够理…
Sharadha Kandasubramanian Sharadha Kandasubramanianis a Senior Technical Program Manager for Amazon Bedrock. She drives cross-functional GenAI programs for Amazon Bedrock, enabling customers to grow and scale their GenAI workloads. Outside of work, she’s an avid runner and biker who loves spending time outdoors in the sun. Gareth ### Gareth Jones Gareth Jones is a Product Manager at Anthropic focused on the Claude API. His work includes collaboration with AWS on Amazon Bedrock to make Claude widely accessible to developers. Loading comments… ### Resources ### Blog Topics ### Follow ## Lea…
Claude 3 Supported and Unsupported Countries Unlike what was before, in 2025, Claude is now available in most parts of the world. This includes US, Canada, the United Kingdom, nearly all of Europe, Australia, Japan, South Korea, other countries across Asia, Africa, and Latin America. Anthropic has made it clear that Claude is not available in mainland China or a few other restricted places due to local laws and security rules. People and companies in supported countries can use Claude through its website, API, or tools like Claude Code (claude code might not be available in your country mis…
bash aws bedrock-runtime invoke-model \ --model-id us.anthropic.claude-opus-4-7 \ --region us-east-1 \ --body '{"anthropic_version":"bedrock-2023-05-31", "messages": [{"role": "user", "content": "Design a distributed architecture on AWS in Python that should support 100k requests per second across multiple geographic regions."}], "max_tokens": 32000}' \ --cli-binary-format raw-in-base64-out \ invoke-model-output.txt
Bash For more intelligent reasoning capability, you can use Adaptive thinking with Claude Opus 4.7, which lets Claude dynamically allocate thinking token budgets based on t…
For Enterprise organizations, Claude Design is off by default. Admins can enable it in Organization settings. Start designing at claude.ai/design. []( ## Related content ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. Read more ### Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors Read more ### Anthropic expands partnership with Google and Broadcom for multiple gigawatts of next-generation compute Read m…
Methodology All evals were run with a 64K thinking budget, interleaved scratchpads, 200K context window, default effort (high), default sampling settings (temperature, top_p), and averaged over 5 independent trials. Exceptions: SWE-bench Verified (no thinking budget) and Terminal Bench (128K thinking budget). Please see the Claude Opus 4.5 system card for full details. []( ## Related content ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters mo…
Context compaction (beta). Long-running conversations and agentic tasks often hit the context window. Context compaction automatically summarizes and replaces older context when the conversation approaches a configurable threshold, letting Claude perform longer tasks without hitting limits. 1M token context (beta). Opus 4.6 is our first Opus-class model with 1M token context. Premium pricing applies for prompts exceeding 200k tokens ($10/$37.50 per million input/output tokens), available only on the Claude Platform. 128k output tokens. Opus 4.6 supports outputs of up to 128k tokens, which let…
Image 21: logo > Claude Opus 4.7 is the most capable model we've tested at Quantium. Evaluated against leading AI models through our proprietary benchmarking solution, the biggest gains showed up where they matter most: reasoning depth, structured problem-framing, and complex technical work. Fewer corrections, faster iterations, and stronger outputs to solve the hardest problems our clients bring us. > > Ben Chan > > Chief AI Officer Image 22: logo [...] Image 13: logo > Claude Opus 4.7 is a very impressive coding model, particularly for its autonomy and more creative reasoning. On CursorBenc…
The changes mark an escalation from previous measures that targeted companies located within China. Dario Amodei, CEO and co-founder of Anthropic. Photo: AP alt=Dario Amodei, CEO and co-founder of Anthropic. Photo: AP> Anthropic, the developer of the Claude AI models known for their coding capabilities, is not the first US AI start-up to restrict access from mainland China and Hong Kong. Last July, OpenAI also tightened its policies to block access from "unsupported countries and territories". Meanwhile, China has implemented a registration system for AI services. Currently, all approved mode…
Related content ### Introducing Claude Design by Anthropic Labs Today, we’re launching Claude Design, a new Anthropic Labs product that lets you collaborate with Claude to create polished visual work like designs, prototypes, slides, one-pagers, and more. ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. ### Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors ### Products ### Models ### Solutions ### Clau…
| Jurisdiction | Claude.ai Access | Reason for Restriction | API Access | --- --- | | Hong Kong | Blocked | Anthropic regional access policy, NSL risk environment | Not supported | | Mainland China | Blocked | Anthropic policy plus domestic AI registration requirements | Not supported | | Taiwan | Available | Supported region | Supported | | Singapore | Available | Supported region | Supported | | Japan | Available | Supported region | Supported | ## How the Block Works Technically [...] Potentially, but with important caveats. Anthropic's September 2025 terms update blocked access for any co…
上班族為什麼愈來愈常選Claude做簡報？問題從這裡開始有趣起來。ChatGPT在文字問答上依然強大，但論到視覺設計感、排版邏輯、把一大段文字轉化成有結構美感的投影片，Claude用戶口碑明顯比前者高出一截。特別是Claude的Artifacts功能——讓模型直接在對話視窗裡生成可預覽的HTML或互動內容——讓許多設計力有限的上班族直接省掉了還要把生成內容轉去其他軟體的麻煩。 延伸閱讀：Claude是什麼？Artifacts怎麼用？ 根據Anthropic官網資訊，4月16日推出的Claude Opus 4.7，是Claude系列目前最高階的模型，Anthropic官方特別強調這一代在視覺美學品味上的提升，聲勢大到還沒正式上線，Adobe和Figma股價就先跌了一波。 Claude也支援繁體中文，包括台灣用語、正式商業文件、以及口語化的說明文字都能流暢處理。若需要確保輸出使用繁體，指令裡明確加上「請使用繁體中文」即可。 ## Claude Opus 4.7適合哪些人用？ Claude Opus 4.7的功能廣，但有幾種人特別會用到它的核心優勢： ### 常做提案簡報的上班族 業務、行銷、專案管理人員，每週需要產出提案、週報、成果報告。Opus 4.7新版Max output上限從64K tokens提升至128K tokens，其實就是一次能生成更長的內容而不被截斷。過去模型往往…
Read more ### Introducing Claude Opus 4.7 Our latest Opus model brings stronger performance across coding, agents, vision, and multi-step tasks, with greater thoroughness and consistency on the work that matters most. Read more ### Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors Read more []( ### Products Claude Claude Code Claude Code Enterprise Claude Code Security Claude Cowork Claude for Chrome Claude for Slack Claude for Excel Claude for PowerPoint Claude for Word Skills Max plan Team plan Enterprise plan Download app Pricing Log in to Claude ### Models…
3.10 Adaptive thinking only Fixed thinking budgets are gone. Use i.j4i.i2
thinking: {type: 'adaptive'}
plus the i.j4i.i2
effort
parameter. Old API calls with i.j4i.i2
budget_tokens
return HTTP 400. Find and replace before flipping the model flag: i.j4i.i2
client = anthropic.Anthropic() response = client.messages.create( model="claude-opus-4-7", max_tokens=16000, thinking={"type": "adaptive"}, messages=[ { "role": "user", "content": "Explain why the sum of two even numbers is always even.", } ], )
If you’re not an engineer, you don’t need to memorize the syntax. Just give your agent this documentation: (remove…
Share US AI giant Anthropic bars Chinese-owned entities The startup, heavily backed by Amazon, is known for its Claude chatbot and positions itself as focused on AI safety and responsible development. Companies based in China, as well as in countries including Russia, North Korea and Iran, are already unable to access Anthropic's commercial services over legal and security concerns. ChatGPT and other products from US competitor OpenAI are also unavailable within China -- spurring the growth of homegrown AI models from Chinese companies such as Alibaba and Baidu. Anthropic said in a statement…
“This update prohibits companies or organisations whose ownership structures subject them to control from jurisdictions where our products are not permitted, like China, regardless of where they operate,” it said. The changes mark an escalation from previous measures that targeted companies located within China. Dario Amodei, CEO and co-founder of Anthropic. Photo: AP Meanwhile, China has implemented a registration system for AI services. Currently, all approved models are domestically developed. The country has a vibrant open-source community for AI models, including those from DeepSeek and…
web browsers, and user-focused workflows like email management. We implemented several protective measures to combat prompt injection attacks, including specialized reinforcement learning training to help the model recognize and avoid these manipulations, and the deployment of detection systems that can halt the model’s execution when a potential injection attempt is identified. These end-to-end defenses improved both models' prompt injection safety scores compared to their performance without safeguards (see table 3.2A). Model Attack prevention score (without safeguards) Attack prevention sc…
Dario Amodei, CEO and co-founder of Anthropic. Photo via AP. ### AI and Autonomy # Anthropic blocks Chinese-owned firms from AI services over security concerns This move marks the first instance of a major U.S.-based AI company denying access to its technology over national security concerns By Joseph Morton Published Anthropic, the San Francisco-based developer of the Claude AI model, has announced it will cease providing artificial intelligence services to organizations majority-owned by Chinese entities. This policy, effective immediately, also extends to subsidiaries operating in other co…
logo AI编程社区 ## 登录社区云 ### AI编程社区 欢迎加入社区 欢迎加入社区 ## 温馨提示：您尚未绑定手机号 为遵守国家网络实名制规定，未绑定将限制内容发布与互动 立即绑定 AI编程社区 Claude AI国内使用完全指南：免费体验Claude 4最强AI编程助手 # Claude AI国内使用完全指南：免费体验Claude 4最强AI编程助手 Claude AI是由Anthropic公司开发的大型语言模型，具有安全、可靠等技术优势。最新版本Claude 4在代码理解、数学推理等方面有显著提升。相比ChatGPT，Claude在上下文长度、中文理解等方面表现更优。国内用户可通过aicodewith.com平台稳定访问Claude AI，享受官方授权接口、中文界面等便利。平台提供Claude 4、Sonnet等全系列模型，支持编程辅助、内容 训练方法，在安全性、可靠性和有用性方面表现突出。 Claude 产品家族包括： Claude Sonnet：平衡性能与成本的主流版本 Claude Opus：最高性能的旗舰版本 Claude Code：专为编程任务优化的版本 Claude Desktop：桌面应用程序 #### 1.2 Claude 4：技术革新的里程碑 最新发布的 claude 4 在多个维度实现了突破性提升： | 能力维度 | Claude 4 | 前代…
Models Opus Sonnet Haiku Log in Claude.ai Claude Console EN This is some text inside of a div block. Log in to ClaudeLog in to Claude Log in to Claude Download appDownload app Download app # AI research and products that put safety at the frontier AIresearchandproductsthat put safety at the frontier AI will have a vast impact on the world. Anthropic is a public benefit corporation dedicated to securing its benefits and mitigating its risks. ## Project Glasswing Securing critical software for the AI era Continue reading Image 1 Read the storyRead the story ## Latest releases ### Claude Opus 4.…
Skip to main contentSkip to footer | October 2024 | Read system card | | Claude Sonnet 3.5 | June 2024 | Read system card | | Claude 3 | March 2024 | Read system card | | Claude 2 | July 2023 | Read system card | []( ### Products Claude Claude Code Claude Code Enterprise Claude Code Security Claude Cowork Claude for Chrome Claude for Slack Claude for Excel Claude for PowerPoint Claude for Word Skills Max plan Team plan Enterprise plan Download app Pricing Log in to Claude ### Models Mythos preview Opus Sonnet Haiku ### Solutions AI agents Code modernization Coding Customer support Education F…
Apr 17, 2026 Product Introducing Claude Design by Anthropic Labs Apr 16, 2026 Product Introducing Claude Opus 4.7 Apr 14, 2026 Announcements Anthropic’s Long-Term Benefit Trust appoints Vas Narasimhan to Board of Directors Apr 6, 2026 Announcements Anthropic expands partnership with Google and Broadcom for multiple gigawatts of next-generation compute Mar 31, 2026 Announcements Australian government and Anthropic sign MOU for AI safety and research Mar 12, 2026 Announcements Anthropic invests $100 million into the Claude Partner Network Mar 11, 2026 Announcements Introducing The Anthropic Ins…
Afghanistan Belarus Central African Republic China Congo, Democratic Republic of Cuba Eritrea Ethiopia Iran Libya Mali Myanmar Nicaragua North Korea Russia Somalia Sudan Syria Venezuela Yemen
I took their list here and diff'ed it with Wikipedia's list of sovereign states. They don't accept the Democratic Republic of Congo, but the Republic of Congo is OK. There are indeed two Congos with ridiculously similar names, and apparently the one that calls itself democratic is the bad one. Anthropic is fine with Taiwan. (They give two lists, one for API and another for Claude.ai, but the list…
Model descriptionClaude Opus 4 and Claude Sonnet 4 are two new hybrid reasoning large language models from Anthropic. They have advanced capabilities in reasoning, visual analysis, computer use, and tool use. They are particularly adept at complex computer coding tasks, which they can productively perform autonomously for sustained periods of time. In general, the capabilities of Claude Opus 4 are stronger than those of Claude Sonnet 4. Benchmarked CapabilitiesSee our Claude 4 Announcement Acceptable UsesSee our Usage Policy Release dateMay 2025 Access SurfacesClaude Opus 4 and Claude Sonnet…
Models Opus Sonnet Haiku ### Solutions AI agents Code modernization Coding Customer support Education Financial services Government Healthcare Life sciences Nonprofits Security ### Claude Platform Overview Developer docs Pricing Marketplace Amazon Bedrock Google Cloud’s Vertex AI Microsoft Foundry Regional compliance Console login ### Resources Blog Claude partner network Connectors Courses Customer stories Engineering at Anthropic Events Plugins Powered by Claude Service partners Community Campus Program Startups program Tutorials Use cases ### Company [...] ### Models Opus Sonnet Haiku…
| State | Type of Restrictions | Duration | Number of Announced Data Centers | --- --- | | Georgia | Temporary ban | Until March 2027 | 340 | | Maryland | Conditional restrictions | Not fixed | 10 | | Michigan | Temporary ban | Not specified | 21 | | New Hampshire | Temporary ban | 1 year | 0 | | New York | Temporary ban | 3 years | 72 | | Oklahoma | Temporary ban | Until Nov 2029 | 34 | | South Carolina | Temporary ban | Until Jan 2028 | 8 | | South Dakota | Temporary ban | 1 year | 6 | | Vermont | Temporary ban | Until July 2030 | 0 | | Virginia | Conditional restrictions | Not fixed | 498…
Mauritius Mexico Micronesia Moldova Monaco Mongolia Montenegro Morocco Mozambique Namibia Nauru Nepal Netherlands New Zealand Niger Nigeria North Macedonia Norway Oman Pakistan Palau Palestine Panama Papua New Guinea Paraguay Peru Philippines Poland Portugal Qatar Romania Rwanda Saint Kitts and Nevis Saint Lucia Saint Vincent and the Grenadines Samoa San Marino São Tomé and Príncipe Saudi Arabia Senegal Serbia Seychelles Sierra Leone Singapore Slovakia Slovenia Solomon Islands South Africa South Korea Spain Sri Lanka Suriname Sweden Switzerland Taiwan Tajikistan Tanzania Thailand Timor-Leste…