Anthropic Launches Claude Sonnet 5: Near-Flagship Performance, More Affordable Pricing

AI company Anthropic today officially launched Claude Sonnet 5, a major upgrade to its mid-range model, replacing the Sonnet 4.6 released in February. Positioned as the "goldilocks" model in the Claude family — between the lightweight Haiku and the flagship Opus — Sonnet 5 is described by the company as "the most autonomous Sonnet model ever." It can not only devise complex plans but also call tools like browsers and terminals, significantly improving its autonomous capabilities to a level previously only seen in larger, more expensive models.

In terms of performance, the gap between Sonnet 5 and the previous-generation flagship Opus 4.8 has narrowed considerably, while its pricing remains more competitive. Starting today, Claude free users and Pro users will use Sonnet 5 by default. Anthropic also announced a limited-time pricing offer valid through August 31, 2026: $2 per million input tokens and $10 per million output tokens. This pricing strategy undercuts comparable models at similar performance levels, likely attracting more developers and enterprise customers.

Looking ahead, the release of Sonnet 5 marks a key step for Anthropic in balancing performance and cost. As model autonomy improves, the barrier to using Claude in scenarios such as code assistance and automated workflows will continue to drop. Meanwhile, Anthropic’s recent moves in response to U.S. export control policies also hint that the global deployment of AI models is facing a more complex regulatory environment. Whether Sonnet 5 can strike a new balance between compliance and innovation remains a development worth watching.