Claude Sonnet 5 represents Anthropic's latest advance in agentic AI capabilities. The new model handles complex planning, browser automation, and terminal operations while maintaining cost efficiency that previously required larger, more expensive models. Performance approaches their Opus 4.8 tier, marking a significant leap over Sonnet 4.6 across reasoning, tool usage, and coding tasks.

Developers can access Claude Sonnet 5 immediately across all subscription tiers. Free and Pro users get it as the default model, while Max, Team, and Enterprise customers can leverage it through Claude Code and the Claude Platform. Introductory pricing at $2 per million input tokens and $10 per million output tokens runs through August 31, 2026, with standard rates at $3/$15 afterward.

Safety improvements distinguish this release. Pre-deployment evaluations show Sonnet 5 exhibits fewer undesirable behaviors than its predecessor, with better resistance to malicious requests and prompt injection attacks. However, cybersecurity capabilities remain limited compared to Opus 4.8—the model cannot develop working exploits and shows only modest improvement over Sonnet 4.6 on dangerous cyber tasks.

Early adopters report meaningful productivity gains. Testers noted the model completes multi-step coding tasks that previously stalled, performs unprompted output verification, and handles complex workflows like Salesforce updates and enterprise announcements autonomously. These improvements position Sonnet 5 as a practical execution layer for software engineering agents.