Powering the agents: Workers AI now runs large models, starting with Kimi K2.5

Cloudflare has taken a significant step forward in its mission to make its Developer Platform the best environment for building and deploying AI agents. By launching Workers AI, the company is now offering frontier-scale open-source models directly on its AI inference platform, starting with Moonshot AI's Kimi K2.5. This move enables developers to run the entire agent lifecycle on a unified platform, optimizing both performance and cost.

For years, Cloudflare has been building the foundational primitives necessary for creating robust agents. These include Durable Objects for state persistence, Workflows for long-running tasks, and Dynamic Workers or Sandbox containers for secure execution. The Agents SDK, designed to simplify agent development, sits atop these primitives. However, these tools only provided the execution environment; the AI model powering the agent remained a separate consideration.

Workers AI now addresses this gap by running large models that are capable of powering agents with high reasoning capabilities and large context windows. The company's focus on optimizing the inference stack has resulted in significant cost reductions for internal agent use cases. This efficiency allows developers to leverage powerful AI models without compromising on performance or budget.

Cloudflare's journey with Kimi K2.5 began as an experiment, but it quickly evolved into a critical component for internal development tools. Engineers at Cloudflare have adopted Kimi as a daily driver for agentic coding tasks within the OpenCode environment. Additionally, the model has been integrated into the company's automated code review pipeline, with a public demonstration available through the Bonk agent on Cloudflare's GitHub repositories.

In production, Kimi K2.5 has proven to be a fast and efficient alternative to larger proprietary models, without sacrificing quality. By bringing this frontier-scale model directly into the Cloudflare Developer Platform, the company is making it possible to run the entire agent lifecycle on a single, unified platform. This integration streamlines development, deployment, and maintenance, allowing developers to focus on building intelligent agents without worrying about the underlying infrastructure.

The launch of Workers AI marks a significant milestone for Cloudflare's vision of creating a comprehensive ecosystem for AI agents. By offering frontier-scale models on its AI inference platform, the company is positioning itself as a leader in the field, providing developers with the tools and infrastructure they need to build and deploy advanced AI applications seamlessly.

In conclusion, Cloudflare's integration of Kimi K2.5 into Workers AI represents a major leap forward in its mission to make its Developer Platform the go-to environment for building and deploying AI agents. With a focus on optimizing performance and reducing costs, the company is enabling developers to leverage powerful AI models without compromising on efficiency. As Cloudflare continues to expand its offerings, it is poised to become a central hub for AI development, offering a unified platform that simplifies the entire agent lifecycle.