In September 2026, teams managing production LLM traffic face a crucial choice among three AI gateways: Cloudflare AI Gateway, Portkey, and Kong AI Gateway. The pricing gap between these solutions can reach over $50,000 annually, according to a detailed comparison by Tech Insider.
TL;DR
- Teams managing LLM traffic need AI gateways to handle caching, rate limits, cost tracking, and fallback routing.
- Cloudflare, Portkey, and Kong offer different AI gateway solutions with varying pricing models and features.
- The choice of gateway depends on specific needs, existing infrastructure, and budget constraints.
What happened
AI gateways have emerged as essential tools for teams running production LLM traffic, addressing issues like scattered API keys, lack of shared caching, and unexpected cost increases. Three prominent solutions are Cloudflare AI Gateway, Portkey, and Kong AI Gateway, each catering to different needs and budgets.
Cloudflare AI Gateway is a free add-on to Cloudflare's edge network, offering basic features like analytics, caching, rate limiting, and request retries. However, it lacks advanced AI-specific tooling and has log volume limitations.
Portkey, an AI-native gateway, went open source in March 2026, offering a free self-hosted build and a hosted pricing model with three tiers. It supports over 1,600 LLMs and provides semantic caching and LLM guardrails.
Kong AI Gateway extends Kong's existing API management control plane to govern LLM traffic. It offers advanced features like token-level tracking, real-time cost analytics, and support for self-hosted models, but comes with a higher price tag.
Why it matters
AI gateways help teams manage LLM traffic more efficiently, reducing costs and improving resilience. They also address governance and security concerns by providing a single point for logging, redaction, and auditing.
The choice of gateway depends on specific requirements, such as the need for advanced AI-specific features, existing infrastructure, and budget. Cloudflare's solution is ideal for teams already using Cloudflare's edge network, while Portkey suits those looking for an AI-native, open-source option. Kong's gateway is best for enterprises already using Kong for API management.
The pricing gap between these solutions can be significant, with Kong's enterprise deployments potentially exceeding $50,000 annually. Teams should carefully consider their needs and budget when choosing an AI gateway.
Key facts
- Cloudflare AI Gateway is free but has log volume limitations and lacks advanced AI-specific features.
- Portkey went open source in March 2026 and supports over 1,600 LLMs. Its hosted pricing model has three tiers, with the Enterprise tier offering custom guardrail hooks and dedicated support.
- Kong AI Gateway is an extension of Kong's existing API management control plane. It offers advanced features like token-level tracking and real-time cost analytics but comes with a higher price tag.
- The pricing gap between these solutions can reach over $50,000 annually for mid-sized enterprise deployments.
- AI gateways help teams manage LLM traffic more efficiently, reducing costs and improving resilience.
- The choice of gateway depends on specific requirements, such as the need for advanced AI-specific features, existing infrastructure, and budget.
Context
The demand for AI gateways is driven by the increasing complexity and cost of managing LLM traffic. As teams juggle multiple model providers for redundancy and cost arbitrage, they need tools to manage caching, rate limits, cost tracking, and fallback routing.
AI gateways also address governance and security concerns by providing a single point for logging, redaction, and auditing. This is particularly important in the context of shadow AI, where developers may directly wire API keys into scripts, losing visibility into data usage.
The AI gateway market is still evolving, with different solutions catering to various needs and budgets. As the market matures, we can expect to see more specialized features and pricing models.