Cloudflare has expanded User Insights, its AI Gateway analytics tool, to classify AI traffic by the kind of work it represents and flag cases where a model may be more capable than a task requires. The update, announced by Cloudflare, adds task categories, conversation turn analysis, and a Potential Savings view, and arrives alongside an Auto Router that is launching in public beta. The company says the features are available for free to AI Gateway users.
The core addition is a model overkill view that identifies conversations where the selected model appears more capable than the task requires. Cloudflare gives the example of simple formatting or summarization requests being sent to a high-capability reasoning model. The view is not a leaderboard and does not automatically recommend a replacement model, according to Cloudflare. Instead, it lets teams compare cost, latency, token usage, and conversation turns before deciding what to change. A separate Potential Savings view highlights requests that may be handled by a faster or less expensive model without compromising output quality.
Task analysis groups conversations into initial categories including coding, research, writing, summarization, and data analysis. Cloudflare says this provides context that a list of model names cannot, and that the answers will vary by team. Turns analysis shows how much back-and-forth different tasks require, so teams can see the full cost of a task rather than just the first request. Cloudflare notes that a long conversation is not necessarily bad, especially for complex work, but a simple task that keeps taking several turns may point to a prompt, model, or workflow issue worth examining.
The Auto Router, now in closed beta, uses conversation trajectory, task category, task complexity, and model-fit signals to route requests automatically while taking cost into account. Cloudflare says it does not simply send every request to the least expensive model; complex coding or research work may still need a more capable model, while simpler tasks may be handled by a faster or cheaper option. The router uses the same signals that power User Insights.
Those signals come from a dedicated Cloudflare Worker that processes eligible AI Gateway logs. The Worker examines the conversation trajectory, including user requests, assistant responses, tool calls, and tool results, and identifies the type of work being performed. It also returns a confidence score and evaluates dimensions such as task complexity, intent ambiguity, stakes, and context dependence. Cloudflare says the current implementation focuses on a small set of easy-to-understand categories rather than trying to infer every detail about a user’s work. The classification is asynchronous, happening after AI Gateway has handled the request rather than while the user waits for a response.
Cloudflare says metadata is stored separately from log bodies, using Durable Objects for metadata and R2 for log bodies. User Insights exposes derived categories and aggregate views and does not turn the dashboard into a raw prompt browser. Retention of underlying log bodies continues to follow the configured AI Gateway logging behavior, so Cloudflare advises teams to review those settings when deciding what to send through the classifier. The company has not said when the Auto Router will move from closed beta to general availability, and the overkill view remains a starting point for investigation rather than an automated recommendation.
