Editorial illustration for Akamai adds AI Inference and Agent Sandbox to complete its full AI stack
Akamai Launches AI Inference and Agent Sandbox
Akamai rolled out two new AI services on Tuesday, aiming to cover the full lifecycle of building and running AI applications rather than just one piece of it. Akamai Inference gives enterprises managed access to 12 open-weight models, from an 8-billion-parameter option up to a 2.8-trillion-parameter one, through APIs compatible with OpenAI's format. The lineup includes NVIDIA Nemotron 3 Ultra, OpenAI's gpt-oss-120b and Google's Gemma 4, plus embedding models built for retrieval-augmented generation. Companies can run these as serverless, pay-per-use endpoints or as dedicated deployments, with Akamai handling orchestration, authentication and connections to storage and databases behind the scenes.
The second piece, Akamai Agent Sandbox, addresses a different problem: AI agents increasingly write and execute their own code, and that code needs somewhere safe to run. The sandbox gives agents an isolated environment for that execution, separate from production systems.
Both services sit on top of Akamai's existing cloud footprint, which the company says includes GPU and CPU clusters in core regions plus more than 4,350 edge locations worldwide. The timing reflects where Akamai's customers have been pushing the company's roadmap.
Akamai Inference provides managed access to open-weight and custom AI models through OpenAI-compatible APIs, while Akamai Agent Sandbox provides an isolated environment for executing code written by AI agents.
Why this matters
Akamai's pitch is consolidation: build, run, and protect AI without leaving its network. For developers, the OpenAI-compatible API on Akamai Inference means switching providers could be a config change rather than a rewrite, and 12 open-weight models gives some room to pick a size that fits a budget or latency target. The Agent Sandbox is the more interesting piece for anyone actually shipping agents. Letting AI-generated code run in an isolated environment close to Akamai's edge network addresses a real problem that has slowed agent adoption: nobody wants autonomous code executing near production systems with no containment.
That said, bundling inference, execution, and security under one vendor creates lock-in by design, even if Akamai frames it as convenience. Founders should weigh egress costs and model selection against the simplicity of one bill and one dashboard. Researchers will want to see benchmarks on the 8-billion-parameter end of that model lineup before trusting it for anything beyond prototyping.
Akamai is betting enterprises want fewer vendors managing their AI stack. Whether that bet pays off depends on how the sandbox performs under real agent workloads, not just the demo.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives Bengaluru Startup Builds AI Agents for Existing Hardware Design Tools
From Archives Anthropic Ramps Up Biology Efforts, Predicts AI Will Cure Most Human Diseases
Common Questions Answered
What open-weight models are available through Akamai Inference?
Akamai Inference provides managed access to 12 open-weight models ranging from an 8-billion-parameter option up to a 2.8-trillion-parameter model. The lineup includes NVIDIA Nemotron 3 Ultra, OpenAI's gpt-oss-120b, Google's Gemma 4, plus embedding models built for retrieval-augmented generation. This diverse range allows enterprises to select models that fit their specific budget and latency requirements.
How does Akamai Inference's API compatibility benefit developers?
Akamai Inference uses OpenAI-compatible APIs, which means developers can switch between providers with just a configuration change rather than requiring a complete code rewrite. This flexibility reduces vendor lock-in and allows teams to experiment with different models and providers without significant development overhead.
What is the purpose of Akamai Agent Sandbox?
Akamai Agent Sandbox provides an isolated environment for executing code written by AI agents, enabling safe deployment of AI-generated code in production environments. This security-focused service allows developers to run agent-generated code close to Akamai's edge network while maintaining isolation and protection.
How does Akamai's full AI stack approach differ from competitors?
Akamai's pitch centers on consolidation, allowing enterprises to build, run, and protect AI applications without leaving its network. Rather than offering just one piece of the AI lifecycle, Akamai Inference and Agent Sandbox together provide a comprehensive solution that covers the entire process from model selection to secure code execution.
Further Reading
- Akamai expands its AI Cloud platform with dedicated inference and agent execution environments - SiliconANGLE
- Akamai Adds Pair of Managed AI Services to Securely Access Custom Models - Techstrong.ai
- Akamai launches AI inference & agent sandbox tools - IT Brief Asia
- Akamai launches AI inference and agent sandbox services on Akamai Cloud - Express Computer
