Anthropic Claude Release History Timeline

Anthropic introduces its first public assistant.

Anthropic introduces Claude and Claude Instant after closed alpha testing with partners including Notion, Quora, and DuckDuckGo.

Claude can work across hundreds of pages.

Anthropic expands Claude's context window from 9K to 100K tokens for long documents and multi-hour conversations.

A stronger model reaches a public chat website.

Claude 2 launches with better coding, math, reasoning, longer responses, API access, and a public beta chat site at claude.ai.

The fast model gets a practical upgrade.

Anthropic releases Claude Instant 1.2 with gains in math, coding, reasoning, safety, multilingual work, and structured responses.

Claude gets its first paid consumer plan.

Anthropic introduces Claude Pro for the Claude.ai chat experience, offering more usage, priority access, and early feature access.

Claude doubles down on enterprise context.

Claude 2.1 adds a 200K token context window, lower hallucination rates, system prompts, and beta tool use.

Anthropic explains how to steer 200K retrieval.

Anthropic publishes guidance for improving retrieval behavior across Claude 2.1's 200K context window.

Opus, Sonnet, and Haiku reset the model lineup.

Anthropic announces Claude 3 Opus, Sonnet, and Haiku with multimodal capabilities and stronger benchmarks across reasoning, coding, and vision.

The fast model joins the Claude 3 family.

Claude 3 Haiku launches as Anthropic's fastest and most affordable model in the Claude 3 generation.

Claude adds a collaborative business tier.

Anthropic launches a Team plan with higher usage, workspace controls, user management, and centralized billing.

Claude gets a native mobile surface.

Anthropic releases the Claude iOS app with synced chat history and photo upload support.

Claude's API tool calling leaves beta.

Tool use becomes generally available across the Claude API, Amazon Bedrock, and Google Cloud's Vertex AI.

A faster frontier model debuts with Artifacts.

Anthropic launches Claude 3.5 Sonnet, outperforming Claude 3 Opus on many evaluations at Sonnet speed and cost, alongside the Artifacts preview.

Claude adds durable workspaces.

Projects become available on Claude.ai for Pro and Team customers, letting users group chats, knowledge, and work context.

Artifacts become shareable objects.

Claude adds publishing, sharing, and remixing for Artifacts within claude.ai.

Claude reaches Android with full multimodal support.

Anthropic launches the Claude Android app with cross-platform chat sync, vision, multilingual processing, and reasoning capabilities.

Repeated context gets cheaper and faster.

Anthropic introduces prompt caching in beta for the Claude API, letting developers reuse prompt content to reduce latency and cost.

Claude's work canvas rolls out broadly.

Artifacts become available for Free, Pro, and Team users, including creation and viewing on iOS and Android.

Claude adds a secure enterprise plan.

Anthropic announces the Claude Enterprise plan with a 500K context window, more usage, native GitHub integration, SSO, role permissions, and admin tools.

API administration gets organized.

Anthropic adds Workspaces to the Developer Console for spend limits, rate limits, API key grouping, usage tracking, and access control.

Claude starts operating computers.

Anthropic releases an upgraded Claude 3.5 Sonnet, announces Claude 3.5 Haiku, and introduces computer use in public beta.

Claude's document platform gets sturdier.

Anthropic adds PDF support for the new Claude 3.5 Sonnet and token counting for Messages API requests.

Anthropic open-sources a connector standard.

Anthropic introduces MCP as an open standard for connecting AI assistants to external tools, repositories, and business systems.

Models, batches, token counts, caching, and PDFs mature.

Anthropic makes the Models API, Message Batches API, Token Counting API, Prompt Caching, and PDF support generally available.

Claude gains source attribution support.

Anthropic launches citations capability in the API and adds support for plain text and custom content documents.

Hybrid reasoning and agentic coding arrive.

Anthropic launches Claude 3.7 Sonnet, its first hybrid reasoning model, and introduces Claude Code as a terminal coding agent research preview.

Claude becomes easier to trial in existing stacks.

Anthropic launches an OpenAI-compatible endpoint so developers can test Claude by changing API key, base URL, and model name.

Power users get a high-usage tier.

Anthropic introduces Max, a plan for users who collaborate with Claude extensively and need higher limits for demanding projects.

Claude connects to apps and performs longer research.

Anthropic introduces Integrations for remote MCP servers, expands Research to search Google Workspace and connected apps, and makes web search global for paid users.

Opus 4 and Sonnet 4 launch for agents and coding.

Anthropic introduces Claude Opus 4 and Claude Sonnet 4, plus Claude Code GA, extended thinking with tools, and new API capabilities for agents.

A frontier safety threshold enters production.

Anthropic activates AI Safety Level 3 deployment and security protections in conjunction with Claude Opus 4.

Opus gets sharper on coding and agents.

Anthropic releases Claude Opus 4.1 with improvements to agentic tasks, real-world coding, reasoning, research, and analysis.

Claude's context window jumps again.

Anthropic launches beta support for a 1M token context window in Claude Sonnet 4 on the Claude API and Amazon Bedrock.

Claude pilots a browser-operating extension.

Anthropic introduces an experimental Chrome extension that lets Claude read, click, and navigate websites alongside users.

Claude can run Python in a sandboxed API tool.

Anthropic releases a Code Execution Tool that lets Claude execute Python code in a secure sandbox.

Claude starts producing editable work documents.

Claude gains the ability to create and edit Excel spreadsheets, PowerPoint decks, documents, and PDFs directly in the Claude app.

Claude starts remembering work context.

Claude adds memory for Team users and incognito chats for Free, Pro, Max, and Team plans.

Sonnet becomes Anthropic's agent flagship.

Anthropic releases Claude Sonnet 4.5 with major gains in coding, agents, computer use, reasoning, and math.

Near-frontier coding gets faster and cheaper.

Anthropic releases Claude Haiku 4.5, a fast model matching Sonnet 4-level coding and computer-use performance at lower cost.

Claude gets reusable workflow packages.

Anthropic launches Agent Skills in beta, letting Claude dynamically load folders of instructions, scripts, and resources for specialized tasks.

Claude memory expands to more personal plans.

Max users gain access to Claude memory, with Pro plan rollout following gradually over the next two weeks.

Claude enters Microsoft's enterprise AI stack.

Anthropic expands its Microsoft partnership, bringing Claude models to Foundry and Claude-powered options into Microsoft 365 Copilot experiences.

Anthropic's top model jumps for coding and agents.

Anthropic launches Claude Opus 4.5 across apps, API, and major cloud platforms, with improvements in coding, agents, computer use, research, slides, and spreadsheets.

Claude's office workflow gets deeper.

Claude for Excel enters beta for Max, Team, and Enterprise users, while context compaction reduces length-limit errors in long conversations.

Claude's agent toolkit broadens.

Skills gain organization-wide management, partner directories, and an open standard, while Claude in Chrome beta expands to all paid subscribers.

Claude Code's agentic pattern moves into desktop work.

Anthropic launches Cowork research preview on Claude Desktop for Max users, bringing Claude Code-style agentic capabilities to local knowledge work.

Claude enters higher-trust personal and enterprise domains.

Claude mobile gains health and fitness data analysis for Pro and Max users in the US, and Enterprise adds a HIPAA-ready option.

Anthropic formalizes its product incubation lane.

Anthropic announces Labs, a team focused on experimental products at the frontier of Claude's capabilities, citing Claude Code, MCP, Skills, Claude in Chrome, and Cowork.

Agentic work becomes less restricted.

Cowork research preview expands to Pro plan users on macOS, and Claude Code access is added to Team plan standard seats.

Opus gains stronger coding, agents, and 1M context beta.

Anthropic launches Claude Opus 4.6 with stronger coding, longer agentic task reliability, better debugging, everyday work gains, and a 1M token context window in beta.

Claude's office apps get a broader footprint.

Claude for PowerPoint launches, Claude for Excel upgrades to Opus 4.6 with native Excel operations, and health data support expands on mobile.

Enterprise no longer requires a sales motion.

Anthropic makes Enterprise plans purchasable directly, with a seat type that includes Claude, Claude Code, and Cowork.

Sonnet gets a full agent and knowledge-work upgrade.

Anthropic launches Claude Sonnet 4.6 with improvements across coding, computer use, long-context reasoning, agent planning, knowledge work, design, and a 1M context beta.

Cowork gains delegation infrastructure.

Anthropic adds a Cowork plugin marketplace, admin controls, recurring and on-demand scheduled tasks, and a unified Customize section.

Claude memory becomes a baseline feature.

Memory from chat history becomes available for all Claude users, including free users.

Claude shares context across office add-ins.

Claude for Excel and Claude for PowerPoint can share full conversation context, add skills support, and connect through LLM gateways.

Claude creates charts, diagrams, and visualizations inline.

Claude gains the ability to create custom charts, diagrams, and other visualizations directly in its responses.

Cowork can be steered from anywhere.

Pro and Max users can access a persistent agent thread through Claude Desktop or mobile to manage tasks in Cowork.

Claude can use your computer directly.

Pro and Max users can give Claude access to computer use in Cowork and Claude Code, letting it open files, run tools, click, type, and navigate.

Mobile Claude renders live app experiences.

Claude for iOS and Android can connect to fully interactive apps for live charts, diagrams, sketches, and shareable assets in conversation.

Anthropic reveals a cyber-capable frontier preview.

Anthropic announces Project Glasswing, a critical-software security initiative powered by Claude Mythos Preview, an unreleased frontier model for defensive cybersecurity partners.

Claude's desktop agent reaches GA.

Claude Cowork becomes generally available on macOS and Windows through Claude Desktop, with analytics, OpenTelemetry, and Enterprise role-based access controls.

Opus improves for hard coding, vision, and long tasks.

Anthropic launches Claude Opus 4.7 with gains in advanced software engineering, complex multi-step tasks, higher-resolution vision, professional work, and cyber safeguards.

Anthropic Labs adds a visual design product.

Anthropic launches Claude Design, an Anthropic Labs product for collaborating with Claude on designs, prototypes, slides, and one-pagers.

Creative-tool connectors expand Claude's reach.

Anthropic releases connectors designed to help creative professionals use Claude alongside the software they already rely on.

Claude governance connects to security tooling.

Anthropic introduces Claude Compliance API integrations with security and compliance tools so teams can govern Claude across products.

Opus improves across coding, agents, reasoning and work tasks.

Anthropic upgrades Claude Opus to 4.8, with improvements over Opus 4.7 in benchmarks, coding, agentic skills, reasoning, and practical knowledge work.

Enterprise admins get more granular Claude controls.

Anthropic adds connector permissions to Enterprise custom roles, letting admins control which connectors and connector tools each role can use.

Claude Enterprise role controls become more granular.

Anthropic extends Enterprise custom roles with admin permissions for areas like billing and privacy, letting members manage specific administrative functions without becoming Owners.

Anthropic broadens defensive security access.

Anthropic expands Project Glasswing to about 150 additional organizations and says it recently released Claude Security for scanning codebases and suggesting patches with public frontier models.

Anthropic launches its first generally available Mythos-class model.

Anthropic launches Claude Fable 5 for general use with safeguards and Claude Mythos 5 for restricted trusted-access use, then suspends access to both models on June 12 following a US government directive.