Akamai launches AI inference and agent sandbox services

Akamai Technologies has expanded its AI Cloud infrastructure with the launch of two new services: Akamai Inference and Akamai Agent Sandbox. These tools are designed to provide enterprises with a secure environment for deploying and executing AI models and autonomous agents.

Akamai Inference offers managed access to various open-weight and custom AI models through OpenAI-compatible APIs. The initial catalog includes models such as NVIDIA Nemotron 3 Ultra, OpenAI gpt-oss-120b, and Google Gemma 4. The service supports both serverless, pay-per-use inference and dedicated deployments, utilizing Akamai’s global network of over 4,350 edge locations to minimize latency.

Akamai Agent Sandbox provides an isolated environment specifically for executing code generated by AI agents. This addresses growing security concerns regarding agentic AI, which can perform autonomous actions like writing code or managing transactions. By providing a controlled execution layer, Akamai aims to help organizations integrate AI into their workflows without compromising system security.