Claude Sonnet 5 launches with agentic capabilities at $2 per million input tokens
Anthropic's Claude Sonnet 5 brings agentic capabilities—browser control, terminal access, tool use—to a mid-tier model priced at $2 per million input tokens through August 31, 2026.
Anthropic released Claude Sonnet 5, a mid-tier model designed for agentic workflows that require sustained tool use, browser control, and multi-step reasoning without constant human intervention. The company says Sonnet 5 approaches Opus 4.8 performance on agentic benchmarks while costing significantly less, making it practical for developers running long-running automation tasks—code refactoring across multiple files, browser-based data collection, terminal orchestration—where token consumption climbs quickly. The model is live across all Claude tiers (Free, Pro, Claude Code) and the API.
Introductory pricing runs $2 per million input tokens and $10 per million output tokens through August 31, 2026, then rises to $3 input / $15 output. Anthropic also emphasizes improved safety in agentic scenarios: Sonnet 5 shows better refusal of harmful requests and stronger resistance to prompt injection compared to Sonnet 4.6—a critical concern when models have direct access to terminals and browsers. The company is positioning Sonnet 5 as the workhorse for the next wave of agentic AI, where models call tools repeatedly, navigate documentation, and execute multi-step plans.
