Anthropic has launched Claude Sonnet 5, a more powerful version of its midsize language model, as competition among leading AI developers increasingly centers on delivering advanced “agentic” capabilities at lower costs.

The company said Claude Sonnet 5 can independently plan tasks, use tools such as web browsers and computer terminals, and execute longer workflows with minimal human supervision, capabilities that until recently were largely reserved for larger and more expensive AI models.

The launch underscores a broader shift in the AI industry, where agentic functionality is rapidly becoming a standard feature rather than a premium differentiator.

Rivals including OpenAI and Google have recently introduced models with similar autonomous capabilities, signaling that the next battleground will be cost efficiency and reliability rather than simply offering agentic features.

According to Anthropic, Claude Sonnet 5 delivers performance approaching that of its higher-end Opus 4.8 model while significantly reducing operating costs.

The model became the default option for free and Pro subscribers on Tuesday and is available across all subscription tiers.

For developers, Claude Sonnet 5 is initially priced at $2 per million input tokens and $10 per million output tokens through August 31. After that, pricing will rise to $3 per million input tokens and $15 per million output tokens.

Even at the higher rates, Anthropic said the model remains less expensive than its Opus 4.8 offering, as well as comparable models from OpenAI and Google, although it is still priced above Google’s Gemini 3.5 Flash.

The company also said Claude Sonnet 5 delivers substantial improvements over the Claude Sonnet 4.6 model introduced in February, particularly in reasoning, software coding, tool usage and other agentic performance benchmarks.