Skip to content

Anthropic Release notes and new features

San Francisco, US · 106 releases · 36 this month

Website RSS

Also onAcquisitions(2 acquisitions)·Paydays(Pay schedule)·TechConf(3 conferences)

Anthropic releases

September 2026 digest
DateRelease
Sep 23Claude Code 2.1.281unverified
Sep 22Claude Opus 5.5 launched, first model in the Claude 5.5 family
Sep 22Claude API beta lets tools be defined inside mid-conversation system messages
Sep 22Fast mode research preview available for Claude Opus 5.5 on the Claude API
Sep 22Claude Code 2.1.280
Sep 22Claude Opus 5.5 breaking changes: thinking can't be disabled, forced tool_choice removed
Sep 22Claude Opus 5.5 launches with a 1M token context window and 128k output
Sep 19Claude Code 2.1.278
Sep 18Claude Code 2.1.276
Sep 18Claude Code 2.1.277
Sep 18Compliance API adds Claude in Chrome session transcripts in beta
Sep 17Claude Code 2.1.274
Sep 17Claude Code 2.1.275
Sep 15Salesforce in Claude plugin (beta) with 37 pre-built sales skills
Sep 15Claude Code 2.1.272
Sep 15Claude Code 2.1.273
Sep 14Claude Code 2.1.271
Sep 14Claude API input_transformations adds thinking_mismatch_allowed entries in beta
Sep 14Messages API adds on-demand conversation compaction in beta
Sep 12Claude Code 2.1.270
Sep 10Smart reports (beta) on Claude Enterprise analyze team usage and costs
Sep 10Claude Managed Agents permission policies add an auto mode
Sep 10ant CLI 1.32.0 adds sessions connect for live Managed Agents sessions
Sep 3Per-message effort changes beta expands to Claude on Google Cloud
Sep 3ant CLI 1.30.0 adds ant apply to manage agents and resources as code
Sep 1Claude Fable 5.1 and Claude Mythos 5.1 launched
Sep 1Claude API adds turn-scoped system messages in beta
Sep 1Claude API thinking.display adds an updates mode in beta
Sep 1Claude Enterprise Admin, Analytics and Compliance API guides add anthropic-version
Sep 1Claude Fable 5.1 and Mythos 5.1 require 30-day data retention
Sep 1Claude Fable 5.1 output carries text watermarks and C2PA Content Credentials
Sep 1Per-message effort changes enter beta on the Claude API
Sep 1Claude Fable 5.1 and Mythos 5.1 reject tool_choice any and tool
Sep 1Claude Fable 5.1 launches alongside Claude Mythos 5.1 with 1M token context
Sep 1Claude Fable 5.1 thinking blocks only replay to the same or newer models
Sep 1Prompt cache reads on Claude Fable 5.1 and Mythos 5.1 drop to $0.25 per MTok
Aug 27Claude Console adds personal and service account API keys
Aug 27Claude SDKs stop sending beta headers for Files and Skills clients
Aug 26Admin API comes to the ant CLI and Claude SDKs under client.beta.organization
Aug 26Compliance API adds Claude Science and Microsoft 365 session transcripts in beta
Aug 26Compliance API session endpoints are out of beta for Cowork and Claude Code
Aug 25Memory in Claude Cowork, editable memory topics, and sensitive topics setting
Aug 20Claude Python SDK v1.0 moves to httpx2 and requires Python 3.10
Aug 20Claude computer use and browser use toolsets arrive on Google Cloud
Aug 19Claude Console session viewer redesigned with timeline and Inspector panel
Aug 19Claude Managed Agents self-hosted sandboxes can attach memory stores
Aug 19Agent Skills and the Skills API are out of beta on the Claude API
Aug 19Claude API launches browser_toolset_20260801 browser use tool
Aug 19Claude Enterprise Admin API user-management endpoints are out of beta
Aug 19Claude Managed Agents can restrict domains for web_search and web_fetch
Aug 19Computer and browser use toolsets support Claude Fable 5, Opus 5 and Sonnet 5
Aug 19Files API is out of beta on the Claude API
Aug 19Claude computer use tool is out of beta as computer_toolset_20260801
Aug 18Claude Console Workbench is renamed playground
Aug 11Claude API adds the anthropic-workspace-id response header
Aug 11Compliance API returns local Cowork and Claude Code session transcripts in beta
Aug 10Claude Sonnet 5 introductory pricing of $2 / $10 per MTok becomes permanent
Aug 7Claude Managed Agents adds inference_geo to control where inference runs
Aug 7Claude Managed Agents sessions can consult an advisor model
Aug 7Claude Managed Agents sessions load skills from GitHub repositories

Summaries of vendors' own notes. Product names and logos belong to their owners; logos via logo.dev.

About and activity

Anthropic, PBC is an American artificial intelligence (AI) public benefit corporation headquartered in San Francisco, California. Its flagship product is Claude, a series of proprietary large language models (LLMs).

Releases per month
Official release-notes pages 10

Weekly: the week's data and AI releases, Tuesday mornings.