Kong vs. Apigee vs. TrueFoundry: The Per-Request Price Gap

Market Position: Big Picture

Kong, Apigee, and TrueFoundry sit at different points of the same growing category: gateways that manage, secure, and route traffic, increasingly including traffic to LLM providers, not just internal APIs. Kong and Apigee are the two dominant general-purpose API management platforms, both now extending into AI gateway functionality; TrueFoundry is a newer, AI-native entrant built from the start around routing to 250+ LLM providers, with API management as a secondary capability rather than the core product.

The finding that should change how any of the three gets budgeted for AI-routing use cases: The three platforms charge for different things, and the unit price of a request differs by up to ten times. Kong Konnect Plus charges $200 for each additional million requests, plus $100 a month for every LLM model routed. Apigee charges $20 per million standard API calls, plus an environment fee per region that is charged whether or not traffic flows. TrueFoundry charges $25 per seat on its Pro plan, with LLM and tool-call requests pooled across the team and billed at $20 per additional 100,000 beyond the allowance, which is also $200 per million. The comparison only works once you know how many requests, models, regions and seats your deployment will have.

Platform Ratings at a Glance

TrueFoundry rates highest of the three despite the smallest review base. Apigee's split across multiple G2 listings is worth knowing before treating any single cited rating as the complete picture, confirm which specific listing a rating or review count refers to.

Kong: Fast and Flexible, With Request and Model Fees That Climb Quickly

Kong Konnect Plus is a pay-as-you-go plan that starts at $25 a month, billed monthly. It charges $25 a month per serverless gateway control plane (up to five), $200 a month per hybrid gateway control plane (up to two) and $500 a month per Dedicated Cloud Gateway control plane (up to two), with data transfer on dedicated gateways at $0.15 per gigabyte. One million requests a month are included, and each additional million costs $200, up to a 10 million monthly maximum. Each unique LLM model proxied by the AI plugins costs $100 a month, up to five. Published APIs in API Management are $20 a month each for the first 10 (two are included) and $10 for each beyond that, up to 20. Kong Gateway open source remains free to self-host for teams willing to run it themselves. Enterprise is a custom annual contract with no public pricing, volume discounts, and a dedicated customer success manager; independent estimates place small Enterprise deployments at $30,000 to $50,000 annually, scaling into six figures for larger implementations.

The mechanism that matters for AI-routing budgets: Request volume and model count both bill separately, and both hit a ceiling on Plus. At 10 million requests a month, request fees alone are $1,800, about 77% of the bill. Five models add $500, and the plan stops at five models and 10 million requests, so a sixth model or heavier traffic forces a move to Enterprise. Kong's technical performance is independently benchmarked strongly, up to 137,850 requests per second on a basic proxy configuration with 3.82 milliseconds added latency at p95, but the cost of routing scales with traffic and model count, not with a flat platform fee.

Apigee: Deep Google Cloud Integration, With Environment Fees That Compound by Region

Apigee prices on two independent dimensions in its pay-as-you-go model: per-call fees of $20 per million standard API proxy calls (dropping to $16 from 50 million and $13 above 500 million) and $100 per million extensible proxy calls, and a separate per-environment, per-region fee of $365 a month at the Base tier, $1,460 at Intermediate and $3,431 at Comprehensive. Base supports up to 50 queries per second and a service-level agreement (SLA) up to 99%; the API Analytics add-on is $20 per million calls and Advanced API Security is $350 per million calls, both available only on Intermediate or Comprehensive. Network charges are billed separately. Pay-as-you-go has no minimum commitment. Subscriptions are sold in Standard, Enterprise and Enterprise Plus tiers with entitlements published (for example, Standard includes up to 1.25 billion standard calls and three environments) but no prices; independent estimates place typical annual spend at $25,000 to $100,000 or more, and some sources cite $250,000 to $500,000 or more for larger multi-region enterprise deployments.

The mechanism that compounds for multi-region or multi-environment deployments: The environment fee is charged per environment, per region, independent of call volume. A team running the same workload across two regions for redundancy or latency pays the environment fee twice, on top of identical call charges, a structural cost that has nothing to do with how much traffic is flowing. Users consistently praise Apigee's Google Cloud integration and analytics depth while citing cost and a steep learning curve as the most common complaints.

TrueFoundry: AI-Native, With Seat Pricing and a Pooled Request Allowance

TrueFoundry prices on seats plus requests. The Developer plan is free for up to three users, with 10,000 requests and tool calls per user a month and seven days of log retention. The Pro plan is $25 per user a month with unlimited users, 20,000 requests and tool calls per user a month pooled across the team, 30 days of log retention, up to 10 teams and 10 budget rules, and built-in guardrails at no charge. Overage is $20 per additional 100,000 requests. Enterprise is a custom annual commitment with virtual private cloud (VPC), on-premises and data-residency options, and third-party guardrails. Model-provider charges from OpenAI, Anthropic, Azure and others are billed separately on every plan. The AI Gateway unifies access to 250+ LLM providers through a single OpenAI-compatible API, with Model Context Protocol (MCP) gateway support for agentic workflows, a capability Kong and Apigee are still extending into rather than having built around from the start.

The structural difference from Kong and Apigee: TrueFoundry meters only LLM requests and MCP tool calls, so ordinary API traffic does not count against its allowance, and it has no per-model or per-environment fee. Seats and traffic are separate, so growing traffic adds overage rather than seats. The cost curve is steepest for AI-heavy workloads: At $200 per million overage requests, TrueFoundry's rate matches Kong's, and ten times Apigee's standard call price. Running it inside your cloud account is an Enterprise option, which brings infrastructure and DevOps work with it.

What Users and Reviewers Report

Kong: The Upside

It has independently benchmarked performance of up to 137,850 requests per second on a basic proxy with 3.82 milliseconds added latency at p95, and the largest review base of the three at about 322 reviews.

Kong Gateway open source is free to self-host, and Konnect Plus starts at $25 a month with usage billed on top.

Kong: The Downside

Requests above the first million cost $200 per million, and each LLM model adds $100 a month; Plus caps out at five models and 10 million requests.

Enterprise pricing is unpublished, with independent estimates of $30,000 to $50,000 for small deployments and six figures beyond.

Apigee: The Upside

Users consistently praise its Google Cloud integration and analytics depth, and it offers API monetization and full-lifecycle governance.

It has a 60-day evaluation sandbox.

Apigee: The Downside

The per-environment, per-region fee compounds independent of traffic and is charged for staging and development environments too.

Cost and a steep learning curve are the most common complaints, and its ratings are split across several separate review listings.

TrueFoundry: The Upside

It is the highest rated of the three at 4.6 out of 5, with an AI gateway to 250 or more LLM providers through one OpenAI-compatible API and MCP Gateway support.

A free Developer plan, flat $25 per-seat Pro pricing with no per-model or per-environment fee, and built-in guardrails at no charge on Pro.

TrueFoundry: The Downside

Heavy LLM traffic is billed at $20 per additional 100,000 requests, and VPC or on-premises deployment is limited to the custom Enterprise plan.

API management is a secondary capability, and it has the smallest review base at about 55.

Pricing (October 2026)

Pricing re-checked October 2026 against vendor pages for Kong Konnect Plus, Apigee pay-as-you-go and TrueFoundry Pro. Enterprise and subscription tiers are quote-only, so those ranges are third-party estimates.

TCO Comparison: 20 API Services + 5 LLM Models Routed, 10M Requests/Month

This scenario models a representative mid-size deployment adding AI-routing to an existing API footprint, the case where each platform's billing mechanism produces meaningfully different real costs. Assumptions: Kong on Konnect Plus with one serverless control plane and no developer-portal publishing; Apigee pay-as-you-go with two Base-tier regions and standard proxies; TrueFoundry Pro with a five-person platform team, counting only LLM and tool-call requests. TrueFoundry is shown at two LLM shares because it does not meter ordinary API traffic.

Kong produces the highest bill in this scenario, and 77% of it is request volume. If the 20 API services are also published through Kong's API Management, add $260 a month. Apigee is the cheapest at this volume because its standard call price is one-tenth of Kong's per-million rate, though the figure excludes network charges and the add-ons, and Base does not support extensible proxies. TrueFoundry swings from the lowest to nearly Kong's level depending on how much of the traffic is LLM calls. None of the three figures includes model-provider charges, which are billed separately.

The same deployment at 1M, 5M and 10M requests a month

Same assumptions as above: 20 API services and 5 routed models, Apigee in two Base regions, and TrueFoundry Pro with five seats where 20% of requests are LLM or tool calls. Ten million requests is the Kong Plus ceiling; beyond it, Kong and Apigee move to negotiated enterprise pricing.

Kong is the second-cheapest at 1M requests, and its bill grows fastest because each added million costs $200 against Apigee's $20. Apigee's cost is mostly fixed environment fees, so it is the flattest curve. TrueFoundry's result depends on the LLM share; at 100% LLM traffic its 10M figure is $2,105, close to Kong's.

Costs the Pricing Table Misses

• Kong's model fee applies identically to a low-traffic LLM integration and a high-traffic one. A rarely used specialized model costs the same $100 a month as your busiest, and Plus stops at five; audit routed models against their fees.
• Kong's request fee is the largest line item at volume. Each additional million requests costs $200, so doubling traffic to the 10 million maximum adds $1,800 a month over the included million.
• Apigee's environment fee is charged whether or not an environment is serving production traffic. Staging and development environments carry the same per-environment fee as production; confirm which environments are necessary before multiplying regions.
• Apigee's add-ons and network charges sit outside the headline rates. API Analytics adds $20 per million calls and Advanced API Security adds $350 per million, and networking is billed on usage.
• TrueFoundry's pooled allowance assumes steady traffic. The 20,000 requests per user are pooled, so a small team with a large LLM workload pays overage fast; model monthly LLM and tool-call volume before choosing seats.
• Self-hosting TrueFoundry is an Enterprise decision. VPC and on-premises deployment sit on the custom plan, and running inside your cloud account puts provisioning, scaling and security patching on your team.
• Kong Enterprise and Apigee subscription ranges are wide because they are negotiated, not published. The $30,000 to $50,000 and $25,000 to $100,000 or more figures in this report are independent estimates; get an itemized quote before budgeting either at enterprise scale.
• None of the three platforms' AI Gateway features are mature to the same point in their product lifecycle. Kong and Apigee are extending established API platforms into AI routing; TrueFoundry built AI routing first. Confirm current feature parity against your LLM provider and MCP requirements rather than assuming all three cover the same ground.

Exit and Switching Costs

Gateways sit in the request path of every service, so leaving one means touching every consumer. Only what the vendors publish is stated as fact.

In our assessment Kong carries the lowest exit cost because the open source gateway shares the same core and can run without a paid plan. Apigee carries the highest because proxies and policies are specific to Apigee and the service is tied to Google Cloud. Confirm export formats and termination terms in the contract for any negotiated tier.

Complexity Comparison

Set the rate cards aside: What does each gateway ask of the team that runs it? Kong asks for traffic and model-count discipline, Apigee asks for environment planning, and TrueFoundry asks for request forecasting and, on Enterprise, infrastructure ownership.

Red Flags

Every finding below is independently documented and material enough to influence a purchasing decision.

The Decision Framework

Choose Kong if: Sub-millisecond API routing performance and a mature plugin ecosystem are the primary requirement, and your request volume and model count stay low enough that the $200-per-million request fee and the five-model cap on Plus don't become the problem.

Choose Apigee if: Deep Google Cloud integration, API monetization, and enterprise-grade analytics are real requirements, and your deployment is single-region or the multi-region environment fee is an acceptable tradeoff for the platform's full-lifecycle governance capability.

Choose TrueFoundry if: AI and LLM routing, not general API management, is the primary use case, your LLM and tool-call volume is predictable enough to forecast against a pooled allowance, and you want a seat-based price with no per-model or per-environment fee.

Everyone: Before comparing headline rates, count the requests, models, environments and seats your deployment will need, then apply each platform's billing mechanism to those counts. The platform with the lowest per-unit rate is not reliably the platform with the lowest real bill once your traffic, model and region mix is factored in.

Key Decision Factors

Design decides what you are buying. Kong is a high-performance API gateway billed by requests, models and control planes, Apigee is a full-lifecycle API management platform billed per environment, region and call, and TrueFoundry is an AI-native gateway billed per seat with pooled requests.

Audience decides which one fits. Kong fits teams that need routing speed and a small AI footprint, Apigee fits Google Cloud shops that need governance and monetization, and TrueFoundry fits teams for whom LLM routing is the primary use and whose request volume is forecastable.

Price depends on the plan and the unit. For 20 API services plus 5 LLM models at 10 million requests a month, monthly cost is about $2,325 on Kong, about $930 on Apigee in two regions at the Base tier and about $505 to $2,105 on TrueFoundry Pro, depending on how much of the traffic is LLM calls.

Choose by situation:

• A routing-performance team with a small AI footprint should choose Kong and audit routed models and request volume against its fees.
• A Google Cloud enterprise should choose Apigee and confirm which environments are necessary before multiplying regions.
• An AI-first team with forecastable LLM traffic should choose TrueFoundry and model request overage before picking a seat count.

The Bottom Line

Kong, Apigee, and TrueFoundry are all credible, well-reviewed platforms solving a related but distinctly different problem, Kong for high-performance API routing with AI Gateway extensions, Apigee for full-lifecycle enterprise API governance with AI Gateway extensions, and TrueFoundry for AI and LLM routing built from the ground up. The real finding this comparison surfaces is that the three price different things: Kong bills requests and models, Apigee bills environments, regions and calls, and TrueFoundry bills seats and pooled LLM requests. Request price differs by up to ten times, $20 per million on Apigee against $200 per million on Kong and TrueFoundry overage. Count your real request, model, region and seat footprint before comparing any two of these three.

Free trials and test-drives. Kong lists a 30-day free trial of Konnect Plus with no credit card, with Plus starting at $25 a month plus usage and Enterprise by quote. TrueFoundry has a free Developer plan for up to 3 users and 10,000 requests per user a month, a Pro plan at $25 per user a month with a free-trial button, and a sandbox with no credit card. Apigee is evaluated through Google Cloud sales. Route one real API or model workload through each and measure gateway cost per million requests, since request volume drives the Kong Plus bill.