Vercel AI Gateway

Offering overview

Built for teams already deploying Next.js and AI SDK apps on its serverless platform, the AI Gateway acts as a single endpoint to models from OpenAI, Anthropic, Google, xAI, Cohere, Z.AI, MiniMax, Deepgram, ElevenLabs and many open-source labs, across text, image, video, audio, embedding and reranking modalities. Requests fail over to the same model on an alternate provider when one degrades, and routing can prefer availability, cost or latency. Operational tooling, paired with the Fluid compute runtime for hosting agents, is the differentiator: a dashboard tracks spend, requests, time-to-first-token and token counts, budgets can cap teams, projects or keys, OIDC tokens replace long-lived secrets, and optional zero-data-retention and provider allowlists enforce compliance. Tokens are billed at upstream list prices with no markup or platform fee, bring-your-own-key usage is free of gateway charges, and every team receives free monthly credits for a subset of models. Add-ons such as team-wide ZDR or allowlists cost $0.10 per 1,000 requests, custom reporting and trace drains are metered separately, and enterprise accounts can pay by invoice.

Technical specifications

Containers

Unknown

Kubernetes

Unknown

Root access

Unknown

API access

Unknown

Autoscaling

Unknown

Persistent processes

Unknown

AI capabilities

Model serving

OpenAI GPT, Anthropic Claude, Google Gemini, xAI Grok, Cohere, Z.AI GLM, MiniMax, Deepgram, ElevenLabs, Poolside, Xiaomi MiMo

AI frameworks

Not stated

Inference endpoints

Unknown

Fine-tuning

Unknown

Vector database

Unknown

Model catalog

Unknown

Monitoring

Not stated

Evaluation

Not stated

Security & compliance

Geography

Provider headquarters are separate from offering infrastructure and availability.

Available plans

Pricing can change. Promotional prices are shown separately from standard prices.

No active plans are recorded for this offering.

Questions & answers

Who is Vercel AI Gateway built for?

It is built for teams deploying Next.js and AI SDK apps on Vercel's serverless platform.

What types of models does the gateway support?

The gateway supports text, image, video, audio, embedding and reranking models from providers including OpenAI, Anthropic, Google, xAI, Cohere and others.

How does provider failover work?

When a provider degrades, requests fail over to the same model on an alternate provider. Routing can prioritize availability, cost or latency.

What spending and privacy controls are included?

Teams can track spend, requests, time-to-first-token and token counts in a dashboard. Budgets can cap teams, projects or keys. OIDC tokens, optional zero-data-retention routing and provider allowlists are also supported.