Claude release timeline

Anthropic's first generally available Mythos-class model, combining frontier capability with stronger safeguards for broad use.

• Fable 5 access was restored globally on July 1
• Stronger than Opus 4.8 on long and complex tasks
• Available on Claude, Claude Code, and the Claude API

The same underlying Mythos-class model as Fable 5, with fewer safeguards in selected areas and access limited to trusted partners.

• Initially deployed through Project Glasswing
• Designed for trusted defensive cybersecurity work
• Not generally available to Claude users

A more reliable Opus collaborator for long-running coding, browser, research, and professional workflows.

• Improved judgment and self-verification
• New dynamic workflows in Claude Code
• Optional fast mode for higher-speed execution

A major step forward in difficult software engineering, high-resolution vision, and polished professional output.

• Stronger performance on complex, long-running coding tasks
• Higher-resolution visual understanding
• More consistent output verification

A full Sonnet upgrade across coding, computer use, long-context reasoning, agent planning, knowledge work, and design.

• 1M-token context window in beta
• Improved instruction following and consistency
• Made Opus-class performance more broadly accessible

An Opus upgrade focused on agentic coding, careful planning, large codebases, research, and complex professional work.

• 1M-token context window in beta
• Longer-running and more reliable agentic work
• Improved code review and debugging

A stronger Opus model for coding, agents, computer use, and demanding office workflows, released at a lower price than its predecessor.

• Improved coding and computer-use performance
• More capable long-horizon agents
• New effort parameter for API users

A fast, economical small model that brought near-frontier coding quality to latency-sensitive applications.

• Optimized for speed and cost efficiency
• Strong fit for chat, customer support, and pair programming
• Available across Claude products and the API

A Sonnet release centered on coding, reasoning, computer use, and longer-running autonomous work.

• Stronger coding and agent capabilities
• Improved computer use
• Launched alongside the Claude Agent SDK

The Claude 4 generation introduced Opus 4 and Sonnet 4 with a focus on coding, advanced reasoning, and AI agents.

• Extended thinking with tool use
• Parallel tool execution
• Improved instruction following and memory workflows

Anthropic's first hybrid reasoning model, allowing users to choose between near-instant answers and extended thinking.

• Visible, controllable extended thinking
• Major gains in coding and front-end development
• Fine-grained thinking budgets through the API

Claude 3 introduced the Haiku, Sonnet, and Opus tiers, along with stronger reasoning and native visual understanding.

• Three model tiers for different workloads
• Vision capabilities across the family
• 200K-token context window at launch

Claude 2.1 expanded the context window and improved accuracy for enterprise and long-document workflows.

• 200K-token context window
• Lower rates of false statements
• Introduced tool use in beta

The second major Claude release expanded access and improved coding, reasoning, and long-context performance.

• 100K-token context window
• Public beta availability in the US and UK
• Improved coding and document analysis

Anthropic introduced Claude, its first publicly announced AI assistant built around helpfulness, honesty, and harmlessness.

• First public Claude release
• Built using Constitutional AI research
• Early access through selected launch partners