This version of Claude should offer more efficient performance at lower costs.
Anthropic has released a new version of Claude that's aimed at handling queries not just from human users, but from agentic AI. The company says its new Sonnet 5 model is capable of handling agentic tasks and delivering performance specs similar to its recent Opus models, but at lower prices.
Starting September 1, Sonnet 5 will cost $3 per million input tokens as a base and $15 per million output tokens. Before that date, the prices will be even lower. In contrast, Opus 4.8 currently costs $4 per million input tokens and $25 per million output tokens. It will be available at those prices in Claude Code and the Claude Platform. Sonnet 5 will also be available for all subscriptions, including as the default model for Anthropic's free and pro tiers of Claude plans.
Agentic AI tools make vastly more queries of a model than a human is even capable of making, and those are largely responsible for instances where customers, particularly enterprise businesses, blow huge amounts of their budget on tokens. We have a more detailed explainer here if you want to get into the weeds about tokens and costs for Claude, because there still are free options for casual users. While this may not prevent all over-spending, Sonnet 5 takes advantage of a new tokenizer to offer greater efficiency to parties that are heavy agentic users.