// vs vercel ai gateway

Vercel AI Gateway alternative

Vercel-hosted gateway bundled with the AI SDK.

The short answer

Use Vercel AI Gateway when

You deploy on Vercel and want the AI SDK + observability tightly integrated.

Use llmrelay when

You want half the API bill and don't need Vercel-native integration.

Head-to-head

Vercel AI Gateway

llmrelay

Pricing on premium models

Zero token markup — $5.00/M input, $25.00/M output on Opus 4.7, same as Anthropic direct. Extras are billed separately: provider allowlist and ZDR at $0.10/1,000 requests each, custom reporting at $0.075/1,000 writes.

$2.50/M input on Opus 4.7 (50% off)

Billing model

Varies (see homepage)

Prepaid credit, no expiry

Model coverage

See homepage

Claude, GPT, Gemini, Fable

Transparency

—

Public patch log, hidden-prompt neutralizer

Where Vercel AI Gateway wins

+Native integration with Vercel deployments
+AI SDK ergonomic API
+Observability built in

Where Vercel AI Gateway loses ground

−Full list price on premium models — $5.00/M input on Opus 4.7
−Requires Vercel account, and some team features bill per 1,000 requests
−Rates mirror upstream providers, so they move whenever the provider moves

Migrating from Vercel AI Gateway?

Both providers speak OpenAI-compatible / Anthropic-compatible. Swap two env vars and your existing app keeps working.

Get API key →

Before you switch from Vercel AI Gateway

The head-to-head detail, what the premium models cost here, and the other gateways people weigh at the same time.