Independent reference. Not affiliated with any vendor mentioned. Quote-only vendors render as quote; we do not infer.
GETGET /guides/llm-gateway-pricing-2026

LLM API gateway pricing 2026

Field guide to LLM gateway pricing in 2026. Five products, three pricing shapes, one rule: provider tokens dominate.

Five products, three pricing shapes

Per-request fixed fee (Portkey Production). Bundled with a non-AI gateway (Kong AI Gateway, Apigee AI). Free at the binary level with provider-token pass-through (Cloudflare AI Gateway, LiteLLM Proxy OSS, self-hosted Portkey).

Cache hit ratio is the lever

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%; the gateway is then a routing and observability play, not a cost-reduction play.

Verified 20 June 2026 Oliver Wakefield-Smith, Founder, Digital Signet we mark quote when a vendor does not publish, and stop.