Identify AI model overuse with User Insights
Cloudflare AI Gateway User Insights now flags model overuse and can auto-route cheaper tasks.
Cloudflare updated User Insights for AI Gateway with context on tasks, model fit, cost, and conversation patterns. A new overkill view shows when a selected model is more capable than the task, and a Potential Savings view estimates where a faster or cheaper model could suffice. Auto Router, launching in public beta, applies those signals without a separate routing rule per workload. The capabilities are free for AI Gateway users and categorize work such as coding, research, writing, summarization, and data analysis.