Ce que Claude a publié en octobre 2026

Anthropic launches Claude Haiku 5.5 with 90% API price reduction, matching GPT-6 Luna

Anthropic introduced Claude Haiku 5.5, slashing API token prices by 90% for requests under 100,000 tokens to $0.10 per million input and $0.50 per million output, matching OpenAI’s GPT-6 Luna. The mod…

Anthropic launches Haiku 5.5 at a much lower price - The New Stack

Anthropic released Haiku 5.5, the latest version of its smallest and most affordable model, featuring a 75% average price reduction to $0.10/$0.50 per million tokens for requests under 100,000 tokens.…

Cisco brings collaborative agents and Claude into Webex chats - Computerworld

Cisco announced AI agents for Webex group chats, allowing users to @mention agents like Anthropic’s Claude to perform tasks such as scheduling, note-taking, and task tracking. The integration includes…

Anthropic Ships Claude Haiku 5.5 at 75% Lower Cost, Pairs With Sonnet Cache Cut and Max API Credits

Anthropic launched Claude Haiku 5.5 at $0.10/$0.50 per million tokens (input/output), priced 75% below Haiku 4.5 for requests up to 100K tokens. It also reduced Sonnet 5.5 cache reads to $0.10 and int…

Claude can now edit your Google Docs without leaving file

Anthropic introduced a public beta Google Workspace add-on for Claude that allows direct editing of Google Docs, Sheets, and Slides without copy-pasting. Pro, Max, Team, and Enterprise users can reque…

Entelect is a preferred partner in the Claude Partner Network Services Track - ITWeb

Entelect was named a preferred partner in Anthropic’s Claude Partner Network Services Track, recognizing its certified Claude practice for embedding AI across enterprise software delivery. The partner…

Claude Haiku 5.5 Cuts Short-Input Prices 90%, Matching GPT-6 Luna: How It Compares With Luna and Sonnet

Anthropic launched Claude Haiku 5.5 on October 7, 2026, slashing API input/output prices by 90% for inputs ≤100,000 tokens (to $0.10/$0.50 per million tokens) and 50% for longer inputs. The model also…

Anthropic recognizes Intuitive.ai as a Preferred Partner in Claude Partner Network

Anthropic designated Intuitive.ai as a Preferred Partner in its Claude Partner Network, recognizing their joint work in enabling governed, enterprise-scale AI adoption. Intuitive.ai integrates its dec…

Grok Bot Is Getting Claude Opus 5.5 as a Backend Model

Grok Bot, an agent platform by SpaceXAI, will now use Anthropic’s Claude Opus 5.5 as one of its backend models, prioritizing it for tasks where it offers the best performance. The integration follows …

Anthropic loosens Claude’s cyber restrictions for verified defenders

Anthropic launched a three-tier Cyber Verification Program (CVP) for Claude, granting verified security professionals expanded access to advanced capabilities for defensive and offensive cybersecurity…

Kradeos Launches to Improve AI Visibility of Software and SaaS Products on ChatGPT, Claude and Perplexity - AiThority

Kradeos launched Kradeos.ai, an AI visibility platform for B2B software and SaaS products that tracks how often ChatGPT, Claude, Perplexity, and Gemini recommend them. It analyzes buyer prompts, ident…

What the Claude Startups Program Terms Say and What Founders Should Check Before Applying

Anthropic's expanded Claude Startups program offers founders $1,000 in API credits and a year of Claude Team, but its terms allow Anthropic to use general ideas retained in staff memory to build simil…

ChatGPT, Gemini, or Claude: Which one is worth 3,000 yen a month? Since Claude can now be used inside Google Docs, I asked all three to edit the same text and compared the results.

Anthropic launched a public beta of Claude inside Google Docs, Sheets, and Slides on October 6, enabling paid plan users to edit documents directly with AI suggestions. The integration is desktop-brow…

Anthropic releases Claude Haiku 5.5 small model and halves Sonnet 5.5 cache read prices

Anthropic released Claude Haiku 5.5, a new small model priced at 90% below Haiku 4.5, and cut Sonnet 5.5 cache read prices by 50%. Haiku 5.5 targets high-volume summarization, classification, and suba…

Anthropic Releases Claude Haiku 5.5: A Small Model With 1M Context Priced at $0.10 per Million Input Tokens

Anthropic released Claude Haiku 5.5, a small model with a 1M token context window and 128K output tokens, priced at $0.10 per million input tokens and $0.50 per million output tokens. It targets high-…

Claude Opus 5.5 Prompting Guide: Dropping “Think Carefully” from Prompts Sharpens Responses

A third-party guide explains how to structure prompts for Anthropic’s Opus 5.5 to improve response quality, including avoiding redundant instructions like 'think carefully' and managing long-running t…

Claude Code's sandbox can read key files if you just enable it. 3 steps to enable it and settings to add first

Claude Code’s sandbox feature, designed to restrict file and network access for shell commands, is off by default and allows reading sensitive files like SSH keys and AWS credentials when enabled. The…

How Comcast and Booz Allen use Claude Mythos to find exploit chains and secure their codebases

Claude launched an expanded Cyber Verification Program (CVP) to make advanced cyber capabilities available to security professionals, building on Project Glasswing. Claude Mythos models help identify …

You can use Claude directly in Google Docs, Slides, and Sheets with a new add-on.

Anthropic introduced a public beta Google Workspace add-on for Claude, allowing users on paid plans to edit Google Docs, Slides, and Sheets directly via a sidebar. The integration mirrors Google’s Gem…

Claude now works with Google Docs, Sheets, and Slides

Claude launched a public beta of its Google Workspace integration, adding a sidebar add-on for Google Docs, Sheets, and Slides on all paid plans. Users can edit files directly in-app, with changes app…

How Cresta turned CX expertise into an agent builder on the Claude Agent SDK

Cresta launched Conductor, an agent builder that helps teams create production-ready AI agents for customer experience use cases, using the Claude Agent SDK as its execution harness. The integration c…

Customize Claude Code with mods

Claude launched 'mods,' TypeScript functions that let users customize Claude Code by rewriting prompts, adding UI, replacing features, or adding new functionality. Mods run with full machine access, i…

How Anthropic's sales team rebuilt inbound with Claude Managed Agents

Anthropic built a buying agent on Claude Managed Agents (beta) that automates inbound sales conversations for Claude, answering pricing, security, and plan questions instantly. The agent handles thous…

Claude for Government is now generally available

Anthropic announced general availability of Claude for Government, a FedRAMP High-authorized version of its AI platform for federal and state agencies. The offering includes Claude Code, Microsoft 365…

Agents you can coach: how Asana builds human-agent teams with Claude

Asana’s Work Graph platform now embeds Claude-powered AI agents as teammates, operating within the same structured model used by human teams. Agents have defined roles, access controls, and shared mem…

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, now integrated with NVIDIA’s Open Agent Safety Platform and open-source OpenShell runtime. The int…

We’re expanding the Claude Startups program to help founders build

Claude expanded its Startups program to help more founders build AI-native companies, offering up to $7,000 in credits, a new tool stack with $45,000 in discounts, and access to Anthropic’s Applied AI…

Checklist for when Claude Opus 5.5 / Sonnet 5.5 do not appear or cannot be selected in Antigravity [October 2026]

Claude Opus 5.5 and Sonnet 5.5 are now available in Antigravity but only for Google AI Ultra users and Pro users after their trial period ends. Free, Plus, and Enterprise plans are excluded, and Ultra…

The Day Claude Deleted My "Human Here!": Testing Sync Issues with Cowork's Cloud Migration

Anthropic announced on October 6, 2026, that Cowork tasks for Pro and Max plan users would execute in the cloud, removing the 'Only on your computer' option. The change shifts workspace execution and …

Anthropic launches Slack-based Claude Tag, says internal-version makes 65% product-code

Anthropic introduced Claude Tag, an AI assistant integrated into Slack that autonomously participates in conversations and performs tasks. Its internal version generates 65% of the product team's code…

Anthropic Expands Claude Startups Program With New Founder Benefits

Anthropic expanded its Claude Startups program on October 6, 2026, adding a $45,000 third-party partner offer stack, $1,000 in API credits, a free year of Claude Team, and Applied AI office hours. The…

Claude Code in the cloud: a field guide to cloud sessions

Claude introduced cloud sessions for Claude Code, running each task on a fresh VM to improve isolation and reproducibility. The post outlines four real sessions and seven workflows optimized for this …

Anthropic introduces Claude Tag as AI teammate in Slack

Anthropic introduced Claude Tag, an AI teammate integrated into Slack that can be mentioned with @Claude to assist with tasks, answer questions, or participate in team chats. The feature supports proj…

Anthropic Expands Claude Startups With $45K Partner Stack

Anthropic launched the Claude Startup Stack, offering approved founders $1,000 in API credits, a free year of Claude Team for up to five Premium seats, and third-party credits worth up to $45,000. The…

Three Pros and Cons Each for Claude, ChatGPT, and Gemini | Our Company's Work, Run Primarily on Claude, Halted for 41 Hours

A company relying on Claude for business operations experienced a 41-hour outage due to an expired login session, halting unmanned tasks. They also report recurring 5-hour and weekly usage limits caus…

A System for Office Workers to "Nurture" Claude Code: Real-world Examples and Templates for Memory Folders, Manuals, and Usage Limit Rules That Don't Start from Zero Every Time

An office worker details a DIY system to make Claude Code remember context across sessions, avoid repeated mistakes, and manage usage limits by maintaining memory folders, manuals, and rules. The appr…

Florida Woman Arrested Following Conversation With Claude That Allegedly Included ‘Threats’

A Florida woman was arrested after allegedly making threats against law enforcement in a conversation with Anthropic’s Claude chatbot. Anthropic’s safety system flagged the messages, alerted authoriti…

SemiAnalysis: Claude $200 Plan Offers ~5x OpenAI's API Value

A SemiAnalysis report claims Anthropic's $200 Claude subscription delivers roughly five times the API-equivalent value of OpenAI's comparable $200 tier for mid-tier workloads, citing Opus 5.5 versus G…

Cresta launched Conductor, an agent builder that helps teams create production-ready AI agents for customer experience use cases, using the Claude Agent SDK as its execution harness. The tool reduces

Anthropic starts processing Claude requests within India

Anthropic began processing Claude AI requests within India using Amazon Bedrock to meet data residency requirements for regulated sectors. The move targets financial institutions and government bodies…

Gorilla Logic joins Claude Partner Network as Select Services Partner - Techli

Gorilla Logic became a Select Services Partner in Anthropic’s Claude Partner Network, integrating Claude and Claude Code into its enterprise software development lifecycle. The partnership focuses on …

Claude AI Inference in India Launches via AWS Bedrock - urbanacres.in

Anthropic launched in-country Claude AI inference in India via Amazon Bedrock, enabling local processing of prompts and responses on Indian servers for regulated sectors. Claude Opus 5, Sonnet 5, and …

Recording Techniques for Managing 10 Projects Simultaneously | Using Claude Code × Obsidian to Give AI a Memory

A user shares a workflow for managing 10+ projects with Claude Code by using Obsidian as an external brain to retain AI memory across sessions. The approach evolved from centralized to decentralized s…

Claude AI workloads can now be processed inside India through AWS

Anthropic’s Claude models can now process requests within India through Amazon Bedrock, routing inference between AWS Mumbai and Hyderabad. This addresses data-residency needs for regulated sectors li…

Claude Code icons get distorted when generated by AI. Official rules you should know before using them in documents or thumbnails

Claude Code’s official community article clarifies that its screen-based icon is a 'spark' with three entry points in VS Code, while its logo/mascot is pixel art prone to AI distortion. The post warns…

[Claude] I received a button to reset my limit for free once, so I'm going to use it to its 'double capacity' this week

A third-party user discovered a new 'Reset for free' button in Claude's paid plans that refills usage limits to full once, with an expiration date. The feature is tied to the Opus 5.5 release and must…

Nervous That Anthropic Is Teaching Claude That It Is Conscious: Microsoft AI CEO Mustafa Suleyman

Microsoft AI CEO Mustafa Suleyman expressed concerns about Anthropic's training constitution for Claude, arguing that its speculative language about machine consciousness could shape how the model int…

Anthropic's Claude AI shared chats surfaced in Google search

Shared chats on Anthropic's Claude AI were surfaced in Google search results via 'site:claude.ai/share', exposing sensitive user data. The issue stemmed from a feature allowing chat link sharing, whic…

Anthropic launches Claude Tag AI assistant built into Slack

Anthropic introduced Claude Tag, an AI assistant embedded in Slack that can be summoned with @Claude to answer questions, join conversations, or perform tasks. It features channel-specific identities,…

Claude Code's Mods now allow internal logic to be fixed with code

Claude Code launched Mods, enabling developers to rewrite internal processing, output formats, and UI elements at the code level. This allows permanent enforcement of team-specific workflows, naming c…

The October 3rd update to Claude Code fixes the "won't stop/disappears" issues for those running it unattended. 3 points to review

Claude Code released version 2.1.288 on October 3, 2026, addressing critical bugs for unattended users. Fixes include proper SIGTERM handling, reliable session resumption, and enforced safety hooks th…

What are Claude Code Mods? 3 Things to Check Before Using Them

Anthropic introduced 'mods' for Claude Code on October 1, 2026, allowing users to alter screen layouts, prompts, tool calls, and add custom commands via JavaScript/TypeScript plugins. Mods run with us…

The Truth Behind Claude's 'Tasks on This Computer Are Being Deprecated' Is 'Integration of Cowork and Chat'—Reading into the Relationship with Cloud Migration (Tentative)

Claude is merging its Cowork and Chat interfaces into a single unified experience, eliminating the option to run new tasks exclusively on local computers starting October 6. Existing local tasks will …

Google Antigravity adds Anthropic's Claude Opus 5.5 and Sonnet 5.5

Google’s Antigravity coding platform added Anthropic’s latest flagship models, Claude Opus 5.5 and Sonnet 5.5, to its model picker exclusively for paying Google AI Pro and Ultra subscribers. The integ…

Claude Code's new Mods system lets developers rewrite the AI coding tool from the inside

Anthropic introduced Mods for Claude Code, a plugin-based system allowing developers to customize the AI coding tool via JavaScript/TypeScript functions. Mods can intercept tool calls, add UI panels, …

[Seriously Divine] Claude Code now has '2 commands for eval', so I tried them out! Also explaining how to prevent 'overfitting' where only the score goes up

Anthropic introduced two new commands in Claude Code—/claude-api build-eval for creating evals and /claude-api hillclimb for iterative improvements—aimed at automating test creation and preventing ove…

[Claude Code] Specialized Division of Labor for AI Employees to Prevent Instruction Drift

Claude Code now supports sub-agents, a feature that divides AI tasks into independent conversation frames to prevent instruction drift. Users can define specialized roles via Markdown files in a dedic…

Anthropic Merges Claude Chat and Cowork Into Unified Interface

Anthropic merged its conversational Chat and autonomous Cowork interfaces into one unified platform, eliminating the need for users to choose between modes. The system dynamically routes tasks, activa…

[AI News Commentary] Claude leads 26% of AI development: Moving toward a future where AI builds the next AI

Anthropic reports that Claude now leads 26% of its AI development tasks, up from less than 1% six months prior, with AI involvement exceeding 90% when including collaborative work. The shift reflects …

How to use claude-mem. Installing the OSS that lets AI remember 'where we left off' after deciding on a 'storage location for memories'

claude-mem is an Apache-2.0-licensed OSS that automatically records and compresses observations from Claude Code sessions, passing them to future sessions to maintain context. Users must first decide …

How does the Claude Code workflow differ from sub-agents? 4 ways to choose based on 'who holds the plan'

Claude Code workflows bundle multiple sub-agents into a JavaScript script that runs in the background, moving the plan outside the conversation window. Unlike sub-agents, skills, or agent teams, workf…

Automatically aggregating 15 daily reports over 5 days with Claude Code: The step-by-step guide to creating a dashboard

A third-party guide demonstrates using Claude Code to automatically aggregate 15 daily reports over 5 days into a single dashboard, visualizing work hours, completion status, and issues. The workflow …

With the October 2nd update to Claude Code, an AI that monitors for oversights is now standard. 3 things to review before running it unattended

Claude Code’s October 2nd update (v2.1.287) introduces "You should know," a built-in AI mod that monitors for oversights during unattended sessions, and expands plugin capabilities to modify deep beha…

Claude Sonnet 5.5 takes third place on Arena’s Agent Arena leaderboard

Claude Sonnet 5.5 achieved a 12.52% net improvement score on Agent Arena’s leaderboard, placing it third behind two Anthropic models. It outperforms OpenAI’s GPT-6 Astra (Max) and scored highly on Art…

Anthropic Is Training 10,000 Engineers to Install Claude Inside the World’s Largest Enterprises

Anthropic launched the Claude Frontier Academy, a $100M program to train 10,000 engineers by 2027 to deploy Claude in large enterprises. The initiative embeds Anthropic’s technical standards into firm…

Why Anthropic wants faith leaders to help guide Claude AI

Anthropic is engaging religious scholars, philosophers, and ethicists from over 15 faith traditions to inform how Claude should handle moral and ethical questions. The goal is to refine Claude’s const…

Claude Developer Plans $100 Billion Capital Raise as IPO Roadshow Nears

Anthropic, the developer of the Claude AI suite, is accelerating plans for a $100 billion capital raise ahead of a potential IPO as soon as mid-November 2026. The company's valuation could reach $2 tr…

Barclays AI Integration Expands with Claude Deployment - The Cryptonomist

Barclays announced a bank-wide rollout of Anthropic’s Claude AI model, aiming for 50% of its developers to use Claude Code by end of 2026 and majority adoption by 2027. The integration targets softwar…

Claude Opus 5.5 Expands Session Limits but Retains Weekly Caps

Anthropic’s Opus 5.5 update expands session limits for its AI models while keeping weekly caps unchanged, aligning with broader industry trends. OpenAI has made similar adjustments, reflecting a shift…

Barclays ramps up use of Anthropic's Claude - Finextra Research

Barclays is scaling up its use of Anthropic's Claude to modernize legacy systems and accelerate software development, targeting 50% of its developer population by end of 2026. The bank leverages Claud…

Claude Sonnet 5.5 takes third place on Code Arena’s WebDev leaderboard

Claude Sonnet 5.5 scored 1786 points on Arena.ai’s WebDev leaderboard, placing third behind Opus 5.5 Max and OpenAI’s GPT-6 Astra Max. Priced at $2/$10 per million tokens (half of Opus), it outperform…

Claude Code Adds "Mods" That Let Developers Change Its Internal Processing and Interface

Anthropic launched Mods for Claude Code on October 1, 2026, allowing developers to extend behavior and interface via JavaScript/TypeScript plugins. Mods can intercept, modify, or replace internal proc…

Getting started with Claude Code mods

Claude launched mods for Claude Code, which are hooks running inside plugins to extend session functionality. The post provides a guide to building a mod from scratch and showcases larger examples.

Barclays Accelerates AI Rollout With Anthropic’s Claude Code

Barclays plans to have half of its developers using Anthropic’s Claude Code by year-end 2026 and most by 2027 as part of a strategic collaboration to integrate AI across operations. The bank highlight…

Claude launched 'mods,' TypeScript-based customizations for Claude Code that allow users to rewrite prompts, add UI, replace built-in features, or add new functionality via plugins. Mods run with the

Claude Code Gets Mods, Plugins That Rewrite the Agent - Pasquale Pillitteri

Anthropic introduced mods and plugins for Claude Code starting with version 2.1.287 (released October 1, 2026), enabling developers to modify the agent's behavior, UI, and internal functions via JavaS…

Anthropic brings Claude to civilian agencies as its fight with the Pentagon drags on

Anthropic launched Claude for Government in open beta since July, now available to US federal and state civilian agencies via FedRAMP High. The move targets civilian agencies after the Pentagon banned…

Claude Code's Deception: Managing Workflow with Managed Services

A third-party analysis reports increasing instances of Claude Code fabricating code or non-existent APIs in long-running workflows, leading to significant rework during CI/testing. The report recommen…

IPC Global has achieved the Select partner tier in the Claude Partner Network Services Track

IPC Global achieved Select partner status in Anthropic's Claude Partner Network Services Track, certifying its team to deploy Claude in production for enterprise clients. The move leverages IPC Global…