Claude Sonnet 5: Anthropic Ships a More Agentic Sonnet With Intro Pricing Through Aug 31

Anthropic has introduced Claude Sonnet 5, positioning it as its “most agentic Sonnet model yet,” with stronger planning and tool use (including browsers and terminals) than prior Sonnet releases. Anthropic announcement

What’s new

• More agentic Sonnet behavior: Sonnet 5 is designed to plan and use tools more effectively for longer-running, autonomous workflows. Anthropic announcement
• Broad availability: Anthropic says Sonnet 5 is available “from today” across all Claude plans (default for Free and Pro; also available to Max, Team, and Enterprise), plus Claude Code and the Claude Platform / API. Anthropic announcement
• Introductory API pricing (time-limited): Anthropic lists introductory pricing for Sonnet 5 through August 31, 2026 of $2 / million input tokens and $10 / million output tokens. Anthropic announcement
• Standard API pricing (after the intro window): After August 31, Anthropic says pricing becomes $3 / million input tokens and $15 / million output tokens. Anthropic announcement
• Model name for developers: Anthropic states developers can call the model as claude-sonnet-5 via the Claude API. Anthropic announcement

Developer notes (prompting + migration)

Anthropic’s prompting guide for Sonnet 5 highlights several API behavior changes and tuning levers, including:

• Effort levels as a primary control for trading off cost/latency vs. capability (low, medium, high, xhigh, max). Claude Platform docs
• Adaptive thinking is on by default (and manual “extended thinking” budgeting is removed for Sonnet 5). Claude Platform docs
• Sampling parameters like temperature, top_p, and top_k are not accepted for Sonnet 5 (and return 400 errors). Claude Platform docs

Sources

• Anthropic: https://www.anthropic.com/news/claude-sonnet-5
• Claude Platform docs: https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-sonnet-5