A10 Networks’ new A10 AI Gateway aims to centralize enterprise AI routing, costs and governance
The latest solution from A10 Networks has been revealed, and it is the A10 AI Gateway, a new control plane designed to help organizations manage how AI agents, applications, and large language models (LLMs) are used across their environments.
As enterprises increasingly rely on multiple AI providers and models, keeping track of which models are being used, how much individual workloads cost and whether appropriate access and security controls are in place can become difficult. A10 AI Gateway is designed to bring those functions together through a centralized management layer.
One of its key capabilities is smart routing, which evaluates the type and complexity of an AI request before determining which model should handle it. Simpler workloads can be directed toward more cost-efficient models, while requests requiring advanced reasoning can be routed to more capable models.
The platform also provides identity-based AI access, synchronizing with existing local directory systems so organizations can apply different AI access and routing policies to specific users or groups. For example, engineering and finance teams can be assigned different models and usage policies based on their requirements.
On the cost-management side, A10 AI Gateway provides real-time, per-request cost tracking in dollars, alongside token budgets that can be managed by model or team. Organizations can also establish hard limits, configure soft-limit alerts and control usage through tokens-per-minute (TPM) and requests-per-minute (RPM) settings.
Security is another part of the platform, with A10 AI Gateway designed to work alongside A10’s existing AI security portfolio, including TrojAI, A10 AI Firewall and ThreatX. The company positions the combination as covering different stages of the AI lifecycle, from pre-deployment testing and runtime protection to network and API-level controls.
In terms of deployment, A10 AI Gateway can run as standalone software or with integrated hardware. It can also operate entirely within a customer’s environment, including on-premises, private cloud and air-gapped deployments. This gives organizations greater control over their AI models, data and policies, particularly where data sovereignty requirements are involved.
A10 AI Gateway is scheduled to become available in Q4 2026.
