Vercel AI Gateway adds priority and flex service tiers

Vercel added service tiers to its AI Gateway: a faster tier for interactive requests, a cheaper one for batch jobs, per its changelog. For teams routing across model providers, that's one less integration to maintain, since the same tier setting applies no matter which model answers the request.

Priority-versus-flex pricing already exists at the provider level; several model providers run their own version of it. What's new is putting one tier abstraction in front of all of them. A developer sets a single config instead of learning each provider's separate priority system.

A chat interface where someone's waiting gets the fast tier. A nightly batch job summarizing documents doesn't need it, so it runs flex and cheaper.

Vercel also added Poolside's Laguna S 2.1 the same week, in a free 256K-context version and a paid 1M-context version.

Vercel's changelog doesn't publish the latency or cost gap between tiers, and I haven't tested it myself. Worth watching whether other gateway products answer with their own uniform tier layer.