Anthropic Launches Claude Sonnet 5: A Cost-Effective Agentic AI Model

Anthropic has unveiled Claude Sonnet 5, an enhanced and cost-effective agentic model that redefines expectations in AI development.

• Capabilities: Claude Sonnet 5 can autonomously perform complex tasks like planning, using tools such as browsers and terminals, and operating with minimal human input—previously achievable only by more expensive models. This functionality aligns with recent releases from OpenAI and Google.
• Market Dynamics: The introduction of Sonnet 5 signals a shift from striving for the best agentic capabilities to achieving these capabilities affordably and reliably. It's more economically feasible than competitors like Opus 4.8, OpenAI’s GPT-5.5, and Google’s Gemini 3.1 Pro.
• Pricing Model: It is available for as low as $2 per million input tokens and $10 per million output tokens until August 31, after which prices adjust slightly upwards.
• Performance Comparison: Despite the lower cost, Sonnet 5's performance closely rivals that of Opus 4.8 in areas like reasoning, coding, and knowledge work.
• User Feedback: Users note improvements in completing tasks autonomously and correcting its output without explicit prompts. Safety enhancements include better handling of misuse, deception, and prompt-injection attacks, although it still trails behind Opus 4.8 in cybersecurity tasks.
• Developer Endorsements: Enthusiastic developers, including Daniel Shepard from Zapier and Fabian Hedin from Lovable, highlight its ability to refuse unsafe requests reliably, underscoring its importance in day-to-day automation.