ToolsJuly 21, 2026via Vercel Blog

Service tiers now available on AI Gateway

Why it matters

Vercel democratizes compute tier selection for AI inference, letting builders optimize for latency vs. cost per request across OpenAI and Gemini without provider lock-in. This shifts economics for cost-sensitive production deployments.

Key signals

  • Service tiering live on AI Gateway for OpenAI and Gemini models
  • Three tiers: default (baseline), priority (~1.8-2x cost, faster), flex (~0.5x cost, higher latency)
  • Works across Chat Completions, Anthropic Messages, OpenAI Responses APIs
  • Billing adjusts per-request based on actual tier applied; fallback to default tier at default rate
  • Tier selection via serviceTier under providerOptions.gateway, unified across all supported providers
  • Applied tier returned in provider metadata for request-level confirmation

The hook

Vercel's AI Gateway now lets you pick your price-speed tradeoff. Priority tier costs 1.8-2x more. Flex tier cuts costs in half.

AI Gateway now supports service tiering. Service tiers let you optimize for latency, throughput, and cost per request to match your use case. Pick a faster tier for interactive workloads (less queueing, higher token throughput), or a lower cost tier for background jobs that can tolerate more latency

The week's key stories, every Friday.

For practitioners and enthusiasts — free, in your inbox.

Free forever. No spam.