GitHub Copilot
AI coding assistance and agents for understanding repositories, writing software and reviewing changes across development workflows.
Vercel AI Gateway
vercel.com
Adds Parallel Decisions Fast for web-grounded classification and scoring with probabilities and cited evidence, free through November 23, 2026.
Visit website
Pricing
Freemium
Available on
API, AI SDK, Local development, External servers, Cloud environments, Vercel
Category
Developer Tools
What is Vercel AI Gateway?
Vercel AI Gateway gives developers unified access to AI models through one API key. It provides infrastructure for applications and supported coding agents.
Applications can run locally, on external servers or in cloud environments. Supported interfaces include AI SDK, OpenAI Chat Completions, OpenAI Responses and Anthropic Messages.
Key features
Provider routing and fallbacks
Set provider preferences and fallback models for inference requests.
Request observability
Inspect provider attempts, status, latency, token usage and cost in logs.
Scoped spending budgets
Configure team, project, API-key or member budgets. Enforcement uses soft caps, so spending can exceed a configured threshold before further requests are blocked.
Coding-agent setup
Vercel CLI previews configuration changes and connects selected supported coding agents.
Muse Ember model access
The meta/muse-ember model supports repeated agent subtasks, text, image and video input, tool calling, a one-million-token context window and adjustable reasoning effort. These capabilities are model-specific.
Who is it for?
Application developers comparing models
Connect an application to several model options through a shared integration.
Platform teams monitoring AI operations
Review routing attempts, consumption and costs across application requests.
Agent builders assigning bounded tasks
Evaluate Muse Ember for repeated coding subtasks or document analysis within a larger workflow.
How to use Vercel AI Gateway
- Create a Vercel account and select the team that will own the integration.
- Add a valid payment method to use free credits, or purchase credits for paid access.
- Create a team-scoped API key and store it securely; its value is shown only when created.
- Choose an accessible model and send a request using cURL, TypeScript, Python or a supported client.
- Inspect the request in AI Gateway Logs before configuring routing and spending controls.
Vercel AI Gateway pricing
Freemium
Eligible teams can start with free credits. Paid access uses purchased credits, with token rates matching provider list prices and zero markup. Optional add-ons can incur additional charges, and BYOK fallback requests can consume Gateway credits.
Check current pricing ↗What to know before you use it
- Free credits cover only eligible models, have lower per-model rate limits and exclude Bring Your Own Key. Purchasing credits ends recurring free credits.
- Budgets are soft caps, rather than guaranteed zero-overshoot limits.
- Bring Your Own Key spend is metered separately and excluded from Gateway budgets. Exceeding a scoped budget nevertheless blocks requests in that scope, including BYOK requests.
- Failed BYOK requests can retry with system credentials and consume Gateway credits. BYOK therefore does not guarantee exclusive use of your provider account.
Frequently asked questions
Does my application need to run on Vercel?
No. External servers, clouds and local environments can authenticate with an API key. Vercel deployments can alternatively use OIDC.
What is required to try the free tier?
You need a Vercel account, a team-scoped API key and a valid payment method. Free credits apply only to eligible models.
Can I connect an existing coding agent?
Yes, if the agent is supported. Vercel CLI lets you select agents and preview configuration changes before connecting them.