Independent reference. Not affiliated with any vendor mentioned. Quote-only vendors render as quote; we do not infer.
POSTPOST /calculator/llm-gateway

LLM gateway cost

LLM-aware gateway TCO. Tokens-in, tokens-out, cache-hit percentage, and guardrail overhead.

// LLM gateway model 1M req/mo, 60% semantic cache hit
{
"portkey_production": "$99.00", // Production tier base
"kong_ai_on_konnect_plus": "$250.00", // Konnect Plus floor + AI plugin bundle
"cloudflare_ai_gateway": "$5.00", // Workers paid + AI Gateway free
"provider_tokens_saved_by_cache": "$3,000.00", // 60% of $5,000 token bill
"net_position_vs_no_gateway": "savings $2,750+" // after deducting gateway fee
}

When the cache pays for itself

Semantic caching becomes the dominant economic lever on chat-style workloads with repeating context. On RAG workloads with high context variance, hit ratios tend to land below 20%, and the gateway becomes a routing and observability play rather than a cost-reduction play. Kong AI Gateway, Portkey, and Cloudflare AI Gateway all publish semantic-cache features.

See /guides/llm-gateway-pricing-2026 for the long-form field guide.