Anthropic has announced the release of Claude Sonnet 5, the latest addition to its Sonnet model family. The company describes it as the most agentic Sonnet model yet, capable of planning, using tools like browsers and terminals, and running autonomously at a level that previously required larger, more expensive models.
Sonnet 5 narrows the performance gap with Opus 4.8, offering substantial improvements over its predecessor, Sonnet 4.6, in reasoning, tool use, coding, and knowledge work. According to Anthropic, the model provides a much wider range of cost-performance options, allowing users to adjust effort levels to balance cost and performance.
The model is available across all plans starting today, including Free, Pro, Max, Team, and Enterprise. It is also accessible via Claude Code and the Claude Platform. Introductory pricing is set at $2 per million input tokens and $10 per million output tokens through August 31, 2026, after which standard pricing of $3 and $15 per million tokens will apply.
Safety assessments indicate that Sonnet 5 has an overall lower rate of undesirable behaviors than Sonnet 4.6, with improvements in refusing malicious requests and resisting prompt injection attacks. However, it shows somewhat higher rates of misaligned behavior compared to Opus 4.8 and Claude Mythos Preview on certain automated audits.
Anthropic notes that Sonnet 5 was not deliberately trained on cybersecurity tasks. While it can perform routine cyber tasks, it shows substantially poorer performance on dangerous cyber skills, such as developing software exploits, compared to Opus 4.8 and Mythos 5. Cyber safeguards are enabled by default.
Early access testers reported that Sonnet 5 finishes complex tasks where previous Sonnet models would stop short, checks its own output without being asked, and offers attractive pricing for agentic work. The full safety and capability assessment is detailed in the Claude Sonnet 5 System Card.