Loading updates…
Claude 在 2026年9月 发布了 77 项受追踪的更新。 主要主题是 feature updates(17)。 活动主要集中在 2026年9月28日 那一周。 亮点:“How Anthropic's sales team rebuilt inbound with Claude Managed Agents”(2026年9月30日)。
信号分类：2026年9月
17
feature updates
8
new integrations
3
product launches
1
content pieces
浏览 Claude 的其他月份
- cryptobriefing.com
Anthropic’s Claude for Government is now generally available
Anthropic’s Claude for Government (C4G), a security-compliant AI model for US federal agencies, is now generally available after a phased rollout starting June 2025. The product includes classified-us…
- creativebloq.com
You can now use ChatGPT and Claude to speed up 3D art in Cinema 4D - creativebloq.com
Maxon introduced native Model Context Protocol (MCP) support in Cinema 4D 2026.4, enabling artists to use third-party AI assistants like Claude and ChatGPT to automate repetitive tasks within Cinema 4…
- memeburn.com
Why Claude Sonnet 5.5 Is Cheaper Without a Price Cut, and When It Isn't
Claude Sonnet 5.5 maintains the same per-token pricing as Sonnet 5 ($2 input, $10 output) but reduces task costs by up to 30% due to fewer tokens and tool calls. Savings are strongest at default/mediu…
- note.com
Claude Sonnet 5.5 vs. Opus 5.5: Which one should you use for daily tasks? I asked for the same report three times each, and the speed was about 1.5 to 1.9 times faster
Anthropic released Claude Sonnet 5.5 as a faster, cheaper complement to Opus 5.5, with tests showing 1.5–1.9x speed improvements in real-world tasks. Sonnet 5.5 defaults to placing key decisions later…
- Featureclaude.com
How Anthropic's sales team rebuilt inbound with Claude Managed Agents
Anthropic built a Claude-powered buying agent on Managed Agents to handle inbound sales inquiries, answering thousands of customer questions daily and guiding prospects to purchase. Leads from the age…
- Launchclaude.com
Claude for Government is now generally available
Anthropic’s Claude for Government is now generally available for federal and state agencies, offering FedRAMP High-authorized access to coding and agentic work capabilities. The platform includes gove…
- unite.ai
Anthropic Makes Claude for Government Generally Available to Agencies
Anthropic made Claude for Government generally available to federal and state agencies on September 30, 2026, exiting a public beta that began in July 2026. The platform delivers Claude’s coding and a…
- investmentnews.com
Zocks debuts Claude plugin with seven skills for financial advisors
Zocks introduced a Claude plugin offering seven prebuilt skills for financial advisors, enabling AI-driven analysis of client conversations and automated document generation. The plugin leverages Zock…
- pasqualepillitteri.it
Anthropic launches claude.dev, its new developer hub - Pasquale Pillitteri
Anthropic launched claude.dev on September 30, 2026, consolidating developer resources like engineering deep dives, Claude Code and API guides, and videos into a single hub. The site includes playful …
- note.com
[2026 Latest] How Claude's Long-Context Expansion Changed the Order Capacity for 'Meeting Minutes' Side Hustles
A side hustler reports that Claude Opus 4.7’s long-context expansion (from ~200k to ~1M tokens) increased their monthly meeting minutes projects from 18 to 27. The update enables one-shot processing o…
- cryptobriefing.com
NEAR AI Cloud integrates Claude Sonnet 5.5 for coding and agents
NEAR AI Cloud added Anthropic’s Claude Sonnet 5.5 to its model lineup via a blockchain-native inference platform, offering a 1M-token context window and 7x performance improvement over Sonnet 5. The i…
- cryptobriefing.com
ExodusPoint partners with Anthropic to bring AI agent Claude deeper into hedge fund operations
Anthropic’s AI agent Claude will be integrated into ExodusPoint Capital Management’s internal Atlas AI platform, targeting investment research, portfolio management, and operational efficiency. Over 8…
- Integrationclaude.com
Agents you can coach: how Asana builds human-agent teams with Claude
Asana’s Work Graph platform now integrates Claude models to power AI agents with defined roles, shared memory, and scoped access within Asana’s task and project structure. Employees use Claude via Asa…
- tech.sportskeeda.com
Is Claude down right now (September 29, 2026)? Outage status explored
All Claude services are down as of September 29, 2026, with the outage starting at 13:46 UTC and affecting over 11,000 users. The official status page shows an ongoing incident with no estimated resol…
- financialpost.com
Claude AI now leads OpenAI in workplace adoption and Canadian professionals are taking notice
Anthropic’s Claude introduced the Claude AI Professional E-Degree, a structured online training program with four courses focused on practical AI skills for workplace productivity. The program is avai…
- note.com
Claude Sonnet 5.5 Arrives: 'Almost Opus' at Half the Price—How Will Usage Change?
Anthropic released Claude Sonnet 5.5 on September 28, 2026, delivering over 30% faster output and up to 30% lower task costs than Sonnet 5 while matching Opus 5.5 benchmarks in most areas. It excels i…
- propakistani.pk
Claude Sonnet 5.5 is Faster, Smarter and Up to 30% Cheaper
Anthropic released Claude Sonnet 5.5, a mid-tier AI model offering 30% faster responses and up to 30% lower per-task costs than Sonnet 5. It excels in coding (70.6% on Terminal-Bench 4.0), visual task…
- note.com
Claude Code vs. Cursor Pricing Comparison: Even at $20 a Month, Which One Is a Better Deal?
Claude Code is bundled into paid Claude plans (Pro at $20/month, Max at $100+), while Cursor charges separately ($20–$200/month). The key difference lies in usage limits: Claude resets limits on a 5-h…
- cryptobriefing.com
Claude Sonnet 5.5 nears Opus performance while cutting task costs up to 30%
Anthropic released Claude Sonnet 5.5, a new model in the Claude 5.5 family, delivering near-Opus 5.5 performance while reducing task costs by up to 30% and increasing speed by over 30%. It excels in c…
- aiweekly.co
Claude Sonnet 5.5 Ranks #2 at 56, With Record Token Burn
Claude Sonnet 5.5 achieved a 56 score on Artificial Analysis's Intelligence Index, placing it second only to Opus 5.5. It set a record for output token usage per task (~193k tokens), driving a 50% hig…
- edtechinnovationhub.com
Anthropic says Claude lesson-planning tools are reaching 68,000 teachers in Ghana
Anthropic announced that Claude-powered lesson-planning tools are now accessible to 68,000 teachers in Ghana through a partnership with PlayLab and Ghana’s Ministry of Education. The company also deta…
- iclarified.com
Anthropic Launches Claude Sonnet 5.5 With Faster Output and Lower Per-Task Costs
Anthropic released Claude Sonnet 5.5, delivering over 30% faster output and up to 30% lower per-task costs than Sonnet 5. It outperforms Sonnet 5 on Terminal-Bench 4.0 (70.6% vs 10.3%) and nearly matc…
- claude.dev
Building with Claude Sonnet 5.5
Claude introduced Sonnet 5.5, positioning it as a more cost-effective alternative to Opus while maintaining strong performance. The post outlines when to choose Sonnet over Opus, provides updated pric…
- unite.ai
Anthropic Releases Claude Sonnet 5.5 at Unchanged Sonnet 5 Pricing
Anthropic released Claude Sonnet 5.5 on September 28, 2026, positioning it as a faster, lower-cost complement to Opus 5.5. It maintains Sonnet 5’s pricing ($2/$10 per million tokens) while delivering …
- Integrationclaude.com
Giving companies more control over their AI agents, with NVIDIA
Anthropic introduced Claude Managed Agents, a suite of APIs for deploying production-grade AI agents with credential vaulting and audit trails, and partnered with NVIDIA to integrate OpenShell for run…
- Featureclaude.dev
Automating eval design and hillclimbing with Claude
Claude introduced new commands in its claude-api skill—build-eval and hillclimb—to automate evaluation design and optimization against benchmarks. These tools aim to reduce self-deception in model eva…
- officechai.com
Anthropic Releases Claude Sonnet 5.5, Beats GPT-6 Sol On Some Benchmarks
Anthropic released Claude Sonnet 5.5, a major upgrade over Sonnet 5 with over 30% faster performance and up to 30% lower costs for most workloads. It outperforms Sonnet 5 on Terminal-Bench 4.0 (70.6% …
- Featureplatform.claude.com
Claude expands API access with new SDKs and quickstart guides
Claude updated its developer documentation to provide step-by-step guides for making API calls using cURL, CLI, Python, TypeScript, C#, and Go. The guides demonstrate how to interact with the Messages…
- Featurecode.claude.com
Claude expands terminal installation guides for Claude Code across macOS, Linux, and Windows
Claude published a comprehensive terminal guide for first-time users of Claude Code, covering step-by-step installation and usage on macOS, Linux, and Windows. The guide includes platform-specific com…
- Featureplatform.claude.com
Claude expands API with beta features for managed agents and stateful sessions
Claude introduced a suite of new API endpoints and beta features, including the Agents API, Sessions API, and Environments API for managed agent infrastructure, alongside a Message Batches API for cos…
- code.claude.com
Claude Code v2.1.283 introduces auto mode and updated permission defaults
Claude updated its quickstart guide to highlight that Claude Code v2.1.283 or later enables auto mode by default in interactive terminal sessions, allowing the AI to edit files and run commands withou…
- Featureplatform.claude.com
Claude Messages API expands with structured content blocks, citations, and prompt caching
Claude’s Messages API now supports structured input messages with text, image, and multi-block content, including new citation types for web search results, document pages, and content blocks. The API…
- Featureplatform.claude.com
Claude expands official tooling with CLI, seven SDKs, and framework integrations
Anthropic introduced the `ant` CLI and seven client SDKs (Python, TypeScript, C#, Go, Java, PHP, Ruby) for the Claude API, alongside framework-specific libraries like Apple Foundation Models and OpenA…
- platform.claude.com
Claude expands API sessions with managed agents, model overrides, and beta features
Claude introduced a new Sessions API endpoint (`POST /v1/sessions`) that enables dynamic agent management, including pinning specific agent versions, overriding model configurations (e.g., model choic…
- Featureplatform.claude.com
Claude launches Managed Agents beta with new CLI, SDKs, and cloud environments
Claude introduced Managed Agents, a beta feature enabling autonomous agent creation with configurable models, system prompts, tools, and environments. Users can deploy agents in Anthropic-managed sand…
- code.claude.com
Claude expands Linux desktop app with Cowork virtualization and broader distro support
Claude’s Linux desktop app (beta) now supports Cowork via QEMU/KVM virtualization, requiring hardware virtualization and specific packages like QEMU, UEFI firmware, and virtiofsd. The app adds Debian …
- Featurecode.claude.com
Claude launches Code tab in desktop app for interactive local coding
Claude added a new Code tab to its desktop app, enabling users to interactively work on local files with direct access, approval-based editing, and diff reviews. The feature requires a Pro, Max, Team,…
- Featureplatform.claude.com
Claude launches managed agents for reusable, versioned agent configurations
Claude introduced managed agents, a feature allowing users to create reusable, versioned agent configurations that bundle model, system prompt, tools, MCP servers, and skills. Agents can be referenced…
- Featurecode.claude.com
Claude launches mobile app for Claude Code with cloud sessions, Remote Control, and Dispatch
Claude introduced mobile support for its Claude Code product via the existing Claude app (iOS/Android), enabling users to start, monitor, and steer cloud sessions, Remote Control, and Dispatch tasks f…
- platform.claude.com
Claude integrates with Amazon Bedrock for AWS-native model access
Claude now offers access to its models (Opus 4.7+, Fable 5.1, Sonnet 5, Haiku 4.5) via Amazon Bedrock, running on AWS-managed infrastructure with zero Anthropic operator access. The integration uses t…
- code.claude.com
Claude launches cloud-based Claude Code sessions for Pro, Max, Team, and Enterprise plans
Claude introduced cloud-based sessions for running Claude Code in the cloud via claude.ai/code, mobile app, desktop app, or terminal, eliminating local setup. Sessions clone a GitHub repository to an …
- Integrationplatform.claude.com
Claude models now available on Google Cloud's Agent Platform with SDK support
Claude models are accessible via Google Cloud's Agent Platform with a near-identical API to Anthropic's Messages API, requiring minor adjustments like moving `model` to the endpoint URL and `anthropic…
- Integrationplatform.claude.com
Claude integrates with Microsoft Foundry for Azure-native API access
Claude models are now accessible via Microsoft Foundry, enabling Azure-native endpoints and authentication. Users can deploy Claude Opus 5.5, Sonnet 5, and other models through Foundry resources, bill…
- Integrationcode.claude.com
Claude expands third-party cloud integrations for enterprise deployment
Claude Code now supports deployment via Amazon Bedrock, Google Cloud’s Agent Platform (Vertex AI), and Microsoft Foundry, joining existing options like Anthropic Console and self-hosted environments. …
- note.com
Have you ever clicked the model name? How to switch Claude models and when to use Haiku [Claude Glossary #6]
Claude introduced Haiku 4.5 as a fast, cost-effective model with a 200K token context window, priced at $1M input/$5M output tokens. The UI now allows real-time model switching via a dropdown menu, wi…
- siliconangle.com
Anthropic debuts Claude Sonnet 5.5 running 30% faster than the previous-generation AI model
Anthropic introduced Claude Sonnet 5.5, a mid-tier AI model 30% faster than its predecessor at lower cost, optimized for document generation, coding, and design tasks. It includes invisible text water…
- note.com
Who is the '40% Cheaper' Claude Opus 5.5 for? A Checklist to Avoid Confusing Monthly Fees with API Costs
Anthropic announced Claude Opus 5.5 on September 22, 2026, with API input/output unit prices reduced by 20% (to $4/$20 per 1M tokens) and a 40% cost reduction claim tied to usage efficiency. The artic…
- note.com
Claude Opus 5.5と、半額のGPT-6が同じ日に出ました。「遅くしよう」から10日後のことです｜2026年9月のAIニュースを実践者目線でまとめ ...
遅くしようと言った2社が、10日後に値下げした新しいモデルを同じ日に並べました。速くて安いAIほど、「止まる場所」はあなたが先に決めておく。今週は、それを考えさせられる1週間でした。 9月12日、Claudeを作っているAnthropicのCEOが「AIの進歩の速度を調整しよう」と書き、OpenAIのAltmanも賛成しました。その10日後の9月22日（米国時間）、その2社が同じ日に新しいモデルを…
- techtarget.itmedia.co.jp
Claude Opus 5.5、GPT-6 Sol／Luna登場 情シスが迫られるモデル使い分け
Anthropicが投入した「Claude Opus 5.5」は、上位モデル「Fable 5.1」に匹敵する性能をより低い運用コストで発揮するという。一方のOpenAIは「GPT-6 Sol」と「GPT-6 Luna」を発表した。Solは高い推論能力を備え、Lunaは処理速度を重視した大量のワークロードを想定している。
- bnext.com.tw
Claude幫自己寫Code！Anthropic兩周讓App快3倍：如何用4撇步，讓AI自己找問題跟驗收？
Anthropic用Claude衝刺兩周，13項核心指標平均快3.1倍、3,000多次修改零事故。關鍵是先建量化基準讓AI驗收指標，工程師把關上線。 叫AI「幫我把這個改快一點」，就像叫人減肥卻不給他體重計，改了一堆，沒人說得出到底有沒有變好。 Anthropic在2026年9月23日公開的一篇工程文章，示範了另一種用法：先給AI一台體重計，再讓它自己想辦法把數字壓下去。 實驗對象是Anthrop…
- bnext.com.tw
換AI不用重教一次！Claude⇄ChatGPT雙向搬家，記憶、對話紀錄、Skills轉移教學一次看
長期使用 AI，轉移 Claude、ChatGPT 等平台往往最費事，因為累積數月的工作背景、寫作偏好、專案脈絡與操作習慣，可能得重新設定，過去花時間建立的使用默契也很難直接快速轉移。 隨著 Claude、ChatGPT 陸續補上資料匯入、匯出與記憶轉移功能，現在轉移 AI 平台，已經有方法把部分記憶、偏好與工作設定一起搬過去，甚至連 Skills 都能轉移。
- toyokeizai.net
チャットAIとの対話だけではもったいない！ ｢Claude Cowork｣で実現するAIに仕事を"任せる"働き方
Anthropicでは、Claude CoworkをはじめとするAIエージェントの設計において、Human-in-the-loop（HITL：人間の関与）という考え方を重視しています。 これは、AIが作業を代行するとしても、重要な判断や責任は人間が担うべきだという考え方です。 Claude Coworkは自律的に作業を進めますが、完全に自由に行動するわけではありません。通常は、まず作業計画を利用者…
- finance.sina.com.cn
Anthropic Claude Sonnet 5.5 模型偷跑，实测碾压 GPT-6 Sol、直逼 Astra
开发者 Rajath Gowda 直言不讳：「Sonnet 5.5 正在成为 OpenAI 的一个大麻烦。如果它真的接近 Opus 5.5，Anthropic 就要把前沿能力变得极度廉价。这才是 OpenAI 最该担心的事。」 ...
- dtf.ru
Claude Opus подписка: где доступна старшая модель и как не выбирать лимит за два дня
Подписка Claude Opus не существует отдельно, доступна в Pro и Max, управление лимитом, советы по экономии бюджета, работа с Aifory Pro.
- 51cto.com
Jev 火了：准确率追平 Claude，成本却只有 1/293
最抓眼的是 TypeSafe 公布的四个 Workflow 测试：Jev 和 Claude Sonnet 5 的平均准确度同为 67.8%，但单次成本只有后者的 1/293，平均耗时则是 0.4 秒对 78.1 秒。
- note.com
1시간 걸리던 영상 편집을 프롬프트 10분 만에 끝낸 이야기[Claude Code]
Canva로 1시간 걸리던 컷 편집 작업을 프롬프트 작성 10분과 최종 확인 15분, 총 25분 만에 끝냈습니다.소재는 3개로 나누어 촬영한 총 30분 분량의 해설 영상입니다. '어', '저기' 같은 추임새, 다시 말하기, 침묵이 가득했습니다.편집은 전부 Claude Code에 맡겼습니다.제가 쓴 시간은 단 25분뿐얼마 전 담당 서비스를 설명하는 총 30분…
- testingcatalog.com
Anthropic may release Claude Sonnet 5.5 within days
A trusted leaker reports Anthropic is preparing to launch Claude Sonnet 5.5 within days, following the recent Opus 5.5 release. Partners received a second checkpoint with improved performance, and pri…
- note.com
Claude's Favorite Fairy Tales
A third-party analysis explores how an AI assistant named Claude discusses fairy tales, highlighting its focus on ethical, societal, and raw interpretations of stories rather than simple moral lessons…
- cryptobriefing.com
Anthropic’s Claude solves nine-loop amplitude challenge in theoretical physics
Anthropic’s Claude AI, using Fable 5.1 on its Claude Science platform, autonomously computed a nine-loop scattering amplitude in theoretical physics—a frontier problem that took human experts years. T…
- Launchclaude.com
Build plugins for Claude
Claude introduced a new developer portal enabling third-party developers to submit, track, and monitor plugins for approval and listing in the Claude directory. Plugins, which package MCP connectors o…
- unite.ai
Anthropic avaa hakemistojen lähetysportaalin Claude Pluginsille
Anthropic avasi hakemistojen lähetysportaalin 25. syyskuuta 2026, tarjoten maksullisilla Claude‑suunnitelmilla oleville kehittäjille tavan lähettää liitännäisiä…
- Featureclaude.com
Claude Tag now supports personal connectors in channels
Claude Tag now allows users to leverage their own connectors for requests made in Slack channels, enabling access to personal tools like calendars, drives, or CRM accounts. Output can be reviewed befo…
- Featureclaude.com
Coding sessions are longer and use more context. Claude Opus 5.5 is built with that in mind.
Claude released Opus 5.5, a new model version priced 40% lower than Opus 5 for token-based billing, with a 20% reduction in input/output token costs and a 60% drop in cached token reads. The update ta…
- Featureclaude.dev
How we made claude.ai 3x faster in two weeks
Claude achieved a 3x speed improvement for claude.ai in a two-week performance sprint, shipping 3,000 changes safely. The effort involved new benchmarks, iterative Slack-thread loops, and strict guard…
- Contentclaude.com
How to prepare for AI-driven code modernization projects
Claude published a guide on preparing for AI-driven code modernization, emphasizing organizational readiness over technical execution. The article details six steps—defining targets, certificates, pro…
- Integrationclaude.com
Claude Marketplace: one place to discover plugins, agents, and services from our partners
Claude introduced the Claude Marketplace, a centralized hub for discovering plugins, agents, and services from partners. The platform benefits customers by providing a single destination to find tools…
- Integrationclaude.com
How CodeRabbit, Power Digital, and ThoughtSpot scale with Snowflake and Vercel on Claude Marketplace
Claude Marketplace now lets Anthropic commitment holders allocate spend to Snowflake and Vercel, enabling faster procurement for teams already building on Claude. Power Digital, ThoughtSpot, and CodeR…
- helpnetsecurity.com
Claude Opus 5.5 cuts costs and adds safeguards for autonomous AI - Help Net Security
Anthropic released Claude Opus 5.5, a major upgrade to its flagship model, featuring lower API costs ($4/$20 per million tokens), 30% faster output, and built-in safeguards for autonomous AI tasks. Th…
- builtin.com
Claude Code vs. Gemini Spark: How Do They Compare? - Built In
Anthropic’s Claude Code operates with direct human oversight, allowing real-time intervention to prevent errors, while Google’s Gemini Spark runs unattended in the background across cloud tools. The a…
- unite.ai
Anthropic julkaisee Claude Opus 5.5:n alemmalla hinnoittelulla ja uusilla suojatoimilla
Anthropic julkaisi Claude Opus 5.5 22. syyskuuta 2026, ensimmäinen malli sen uudessa Claude 5.5 -perheessä. Mallin hinta on 4 $ per miljoona syötettävää tokenia ja 20 $ p…
- Featureclaude.dev
Getting the most out of Opus 5.5 in Claude and Claude Code
Claude published guidance on how to maximize Opus 5.5's performance, including best practices for prompting and steering long runs within its apps and Claude Code. The update focuses on improving user…
- adtmag.com
Claude Code Turns Software Development Into a Management Job - ADTmag
Anthropic introduced Claude Code Projects, a beta feature that lets developers oversee multiple parallel AI coding sessions through a central coordinator. The system divides large tasks into threads, …
- Featureclaude.com
Working at the frontier: How Balyasny Asset Management evaluates and governs Claude Fable 5
Balyasny Asset Management adopted Claude Fable 5 after rigorous testing, achieving an 89.4% success rate on thousands of financial tasks—up from 86.1% with the prior model. The firm uses Fable 5 in ag…
- Featureclaude.com
Projects redesigned: from folder to conversation
Claude introduced a redesigned Projects feature in beta for Claude Code, shifting from folder-based management to conversational goal-driven workflows. Users can now set high-level goals (e.g., reduci…
- Launchclaude.com
Claude Cowork and chat are now one Claude
Claude has merged its Cowork and chat products into a single unified experience, eliminating the need to choose between them. New capabilities include Claude Docs, Slides, and integrated Design tools,…
- Integrationclaude.com
Bringing Salesforce into Claude
Claude introduced a Salesforce integration plugin in beta, enabling sellers to research accounts, prep for calls, review pipelines, and draft CRM updates directly within Claude. The plugin uses 37 ski…
- marktechpost.com
GPT-6 Astra vs GPT-6.1 Sol vs Gemini 4 Argon vs Claude Fable 5.1: Which Frontier Model Fits Which Job
Claude Fable 5.1 launched on September 1, 2026, alongside GPT-6 Astra, GPT-6.1 Sol, and Gemini 4 Argon. While benchmark scores overlap, pricing and access rules differ significantly. Fable 5.1 and Ast…
