Is Anthropic's Sonnet 5 the first agent that won't stall on complex jobs?
anthropic released sonnet 5, a $3/m input model matching opus 4.8’s agent performance at lower cost. this shift suggests the ai wars are moving from who has the smartest model to who has the cheapest autonomous worker.
anthropic unveiled claude sonnet 5, touting performance comparable to its flagship opus 4.8 but at a fraction of the cost, starting at $3 per million input tokens. the model excels at multi-step agentic tasks like software updates and sales operations without human intervention, solving a key friction point where previous models often stalled mid-task. benchmarks indicate sonnet 5 scores 63.2% on agentic coding and slightly edges out opus 4.8 on knowledge work, offering a higher reliability alternative for budget-conscious developers. additionally, anthropic emphasized safety upgrades, citing improved resistance to prompt injections and fewer hallucinations than prior versions.