A10 Launches AI Gateway for Enterprise Model Control
A10 Networks made its AI Gateway generally available on August 13, giving enterprises a central layer for model routing, access policy, usage visibility, and cost controls. The gateway can steer requests according to task complexity and organizational rules while tracking token budgets and request-level spending. A10 positions it as part of a broader on-premises and private-cloud AI security portfolio.
A10 Networks made the A10 AI Gateway generally available on August 13, according to Help Net Security, after outlining the product in an August 3 company post. The gateway is designed to sit between enterprise users, applications, and agents on one side and multiple AI models on the other, centralizing routing, access control, observability, and cost management.
From model access to policy control
A10 says administrators can set identity-based access and routing policies for individual users or groups, synchronized with existing directory systems. Instead of sending every request to a fixed provider, the gateway evaluates the request type and complexity and can route simpler work to a less expensive model while reserving more capable models for demanding tasks.
The company also describes per-request cost tracking, token budgets by model or team, soft-limit alerts, hard limits, and rate controls measured in tokens or requests per minute. Those features target a common operational problem: organizations often add AI models team by team, leaving security and platform owners without one place to see who is using which service, how much it costs, or whether routing policies are being followed.
A broader A10 security stack
A10 distinguishes the gateway from an AI firewall. The gateway determines who or what may reach a model and where a request should go; an AI firewall inspects prompts and responses for threats such as prompt injection or data leakage. A10 says the gateway complements its TrojAI and ThreatX products, spanning testing before deployment, runtime defense, and network or API enforcement.
The company says the gateway can run as software or with integrated hardware inside a customer-controlled environment, including on-premises, private-cloud, and air-gapped deployments. That makes data sovereignty a central part of A10's pitch, though the retrieved material does not provide independent performance, security-effectiveness, or cost benchmarks.
For platform teams, the practical value is the control plane rather than another model: access, routing, budgets, and observability become shared infrastructure. Buyers will still need to validate provider coverage, routing quality, latency, directory integration, and policy enforcement against their own workloads before treating the product claims as established operational results.
Key Points
- 1A10 Networks made the A10 AI Gateway generally available on August 13 after publishing an official product overview on August 3.
- 2The gateway centralizes identity-based model access, intelligent request routing, per-request cost tracking, token budgets, rate limits, and usage visibility.
- 3A10 says the product can run in customer-controlled on-premises, private-cloud, or air-gapped environments, but the retrieved sources do not include independent performance or security benchmarks.
Scoring Rationale
The launch gives enterprise platform teams a new control-plane option for multi-model access, routing, cost, and governance. It is operationally relevant but the retrieved evidence does not establish independent performance or adoption outcomes, so the impact is meaningful rather than industry-shaping.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
