According to TrendForce, Anthropic has officially launched its mid-range flagship model, Claude Sonnet 5, which focuses on powerful Agent autonomy, browser and terminal tool integration, and other automated capabilities. The new model significantly outperforms its predecessor, Sonnet 4.6, in evaluations such as BrowseComp (smart search) and OSWorld-Verified (computer operation). Users can adjust the effort level to achieve performance comparable to the flagship model Opus 4.8.
In terms of security performance, the new model boasts a lower false positive rate compared to Sonnet 4.6 and has enhanced defenses against malicious requests and keyword injection attacks. Although the model was not specifically trained for cybersecurity, its security capabilities are lower than Opus 4.8. However, it comes with a built-in real-time network security protection mechanism. Additionally, the model has upgraded its tokenizer, increasing the number of tokens generated from the same text by 1.0 to 1.35 times compared to the older version. As a result, the introductory pricing is set at $2 per million input tokens and $10 per output token (valid until August 31, 2026, after which the standard prices of $3 and $15 will be reinstated), aiming to make the actual cost of switching models almost equivalent.
The model is now available to all Free, Pro, Max, Team, and Enterprise users, and can be used in Claude Code.
