Akamai launches AI inference service and expands GPU fleet

Akamai has launched Akamai Inference, a cloud AI inference service offering managed access to 12 open-weight models from 8B to 2.8T parameters via OpenAI-compatible APIs. It is available in beta with serverless pay-per-use or dedicated deployments, alongside the new Akamai Agent Sandbox for running agent-generated code.

• 12 open-weight models, including Nemotron 3 Ultra, gpt-oss-120b and Gemma 4
• Serverless pay-per-use and dedicated deployments available
• GPU fleet expanded with Nvidia Blackwell Ultra clusters
• Support for Nvidia Vera Rubin platform planned for next year