Anthropic launches Claude Sonnet 5, a mid-tier model focused on stronger “agentic” performance while reducing the cost of running AI assistants. Multiple outlets report that Sonnet 5 is positioned as a more affordable option than Anthropic’s flagship Opus models and as an alternative to other high-end models from competitors. TechCrunch and Quartz describe Sonnet 5 as delivering improved agent capabilities alongside lower pricing. Quartz reports launch pricing of $2 per million input tokens, undercutting Opus 4.8, while The Next Web and Mashable say Sonnet 5 narrows the performance gap compared with Opus 4.8 for many tasks but costs less than half as much.
MacRumors adds that Sonnet 5 is Anthropic’s most agentic Sonnet model to date, with capabilities including planning, tool use (such as browsers and terminals), and autonomous task completion. It also says Sonnet 5 improves on earlier Sonnet versions in reasoning, tool use, coding, and knowledge work, and includes safety improvements such as better refusal of malicious requests and lower rates of hallucinations and “sycophancy.” Several sources also state Sonnet 5 is available across all plans at launch and serves as the default model for Free and Pro plans.
