Cloudflare Launches Auto Router for AI Gateway

The new public beta tool classifies requests to balance model performance against costs.

Updated on Oct. 26, 2026 in Artificial Intelligence

Isometric editorial illustration of modular data blocks connected by channels, representing automated cloud-based routing decisions. — Cloudflare launched a public beta for its Auto Router, a tool that uses edge GPUs to automatically assign AI tasks to the most cost-efficient models. AI Illustration. Upload story photo >

Live Poll

Do you trust automated routing systems to correctly handle high-stakes business decisions?

Cloudflare has released its Auto Router in public beta through the AI Gateway platform. The system uses edge GPUs to classify incoming requests and assign them to the most efficient model based on complexity and cost.

Why it matters

By automating model selection, the system helps developers optimize performance while maintaining a balance between expected output quality and operational costs.

The system processes requests capped at 8 KiB of text and classifies them across 14 distinct task categories. Models are scored on a scale of 1 to 5 based on request complexity, ambiguity, stakes, and contextual dependence.

The players

Cloudflare

Cloudflare is a global cloud services provider that delivers content delivery network services, DDoS mitigation, and internet security.

The details

AI Gateway filters models into candidate pools and applies constraints before a classification model calculates a utility formula, defined as expected quality minus adaptive cost penalty. The utility allows users to see routing decisions directly in response headers via specific metadata fields.

October 26, 2026: The Auto Router public beta is officially released.

The Tech Race

This release follows the industry trend of shifting from static model calls toward dynamic, intent-aware request routing. It replaces manual model selection, positioning Cloudflare against other infrastructure providers working to commoditize model access for developers.

Developers can now utilize automated routing to streamline their AI applications without manually managing model pools for every query. While currently free during the beta period, users should be aware that the system does not yet support WebSockets.

The takeaway

Developers implementing AI endpoints should evaluate whether their specific traffic patterns benefit from automated quality-to-cost balancing. As the technology matures, expect more granular control over model selection and broader support for diverse communication protocols.