New AI Models Cost More Than Expected | Tech News

Microsoft’s new Anthropic Claude Sonnet 5 model promises smarter AI for developers—but at a steep price. Tests show it can use up to 12x more tokens than Sonnet 4.6 for architecture tasks, with one run gobbling 47x the usual tokens. Even with lower per-token rates, costs spike unpredictably, making budgeting a nightmare. But Sonnet 5 shines in complex code upgrades, nailing version-specific instructions where older models fail—though at a 10x token cost and sometimes 69 million tokens in wild, uncontrolled searches. The takeaway? Newer models win on precision for specific tasks, but come with wild cost and behavior swings—older models may still be smarter bets for stability and budget control. Listen in comfort:Get a discount on a Soli Pillow: http://solipillow.com/discount/dnn. Advertise on DNN:[email protected] This is an automated, high-level news summary based on public reporting.Report issues to [email protected]. View sources & latest updates:https://sources.thednn.ai/fac44f1029f97337