New Model Release: Claude Fable 5 Model Launched

Claude Fable 5 Reasoning Model Core Upgrades

1. Flagship Model for Complex Reasoning & Long-Duration Tasks

• Anthropic’s most capable broadly released model to date, designed for complex reasoning, long-duration agentic tasks, deep research, software engineering, and enterprise knowledge work
• The longer and more complex the task, the more significant its advantage over previous Claude models

2. Enhanced Long-Duration Agent & Complex Engineering Capabilities

• Optimized for multi-stage tasks in Claude Code, Agent Harness, or enterprise automation workflows, including planning, decomposition, tool use, result verification, and continuous iteration
• Can sustain days-long, complex, asynchronous tasks that were previously difficult for prior models to complete reliably

3. Comprehensive Upgrades in Code, Knowledge Work & Visual Understanding

• Strong performance improvements across software engineering, knowledge work, visual understanding, and scientific research
• Significant gains in scenarios such as complex coding tasks, financial document reasoning, chart/table comprehension, and screenshot-to-application recreation

4. Improved Long Context & Large Output Capacity

• Default 1M-token context window
• Supports up to 128k output tokens per request, ideal for long documents, multi-file analysis, complex project contexts, and long-duration workflows

5. Safety & Compliance

• Includes safety classifiers for high-risk domains such as cybersecurity, biology, chemistry, and synthesis; some requests may be rejected or fall back to Claude Opus 4.8
• Not eligible for zero data retention; model traffic requires a 30-day data retention period, which is a key consideration for B2B customers