Cloudflare unifies Workers AI and AI Gateway under one control plane

Image: Primary

Published Monday, August 10, 2026 · 12:06 AM EDT

Cloudflare said it is merging Workers AI and AI Gateway so customers can reach any model provider, including Workers AI, through one path for observability, billing, security, and logging. AI Gateway proxies requests to model providers with logging and access controls, while Workers AI hosts models on Cloudflare-managed GPUs. The company said both already share an AI binding and a unified REST /ai/ endpoint, including a default gateway that adds observability when no custom gateway is set. Cloudflare is piloting unified routing internally and expects further release work in the coming weeks. Existing Workers AI callers can route through a default gateway for request logging, token tracking, and cost attribution; AI Gateway users can call Workers AI models with unified billing and elevated rate limits on Workers AI models.

Sources

The Cloudflare Blog

In this story

Cloudflare

Published by Tech & Business, a media brand covering technology and business. This story was sourced from The Cloudflare Blog and reviewed by the T&B editorial agent team.

Back to Newswire

Keep reading

Full wire

Robotics Capital

GenRobot AI raises $100 million for robot training data infrastructure

Beijing-based GenRobot AI has completed a $100 million Series A+ funding round, TechNode reported. The company develops equipment and software that collect human demonstrations for use in training robots, providing data infrastructure for companies developing robots.

Products

Stripe reports Privy wallet assets up approximately 25x since acquisition

Privy wallet assets have grown approximately 25x and revenue approximately 9x since Stripe acquired the wallet infrastructure company a little over a year ago, Patrick Collison said in an update covered by SaaStr.

AI Infrastructure

Builder runs Kimi K3 on CPU with discontinued Optane memory

A home server builder ran Moonshot AI's 2.8-trillion-parameter Kimi K3 model without a GPU, using discontinued Intel Optane PMem 200 memory and a 40-core Xeon processor, Tom's Hardware reported.

Infrastructure

Blizzard used 3,924 virtual machines for World of Warcraft: Forever launch

Blizzard used 3,924 virtual machines with 70,480 CPU cores and 251,136 GB of memory to support the launch of World of Warcraft: Forever, Tom's Hardware reported. Players and media praised it as one of the company's smoothest launches.

Science

Phase 3 retatrutide trials report weight loss and reduced knee pain

Two Northwestern University-led Phase 3 trials involving more than 1,000 adults found that participants receiving high doses of Eli Lilly's experimental retatrutide lost, on average, up to 25% of their body weight.

AI Products

Apple tests development process giving engineers 50% more coding time

Apple is experimenting with a software development process that gives engineers 50% more time to code and explore while reducing time spent on integration, 9to5Mac reported, citing Bloomberg's Mark Gurman. The experiment comes as coding agents have improved.