Compare · Vercel AI Gateway and Selan
Is there a Vercel AI Gateway alternative for a team's Claude Code?
Yes, if the traffic is your team's coding tools and you want each developer to sign in rather than hold a key. Vercel AI Gateway is a hosted API for apps and coding agents, with 400+ models at the provider's list price and no markup. Claude Code reaches it with an AI Gateway API key per developer, or with the developer's own Claude subscription passed through. Selan is a hosted gateway that each developer signs in to once. If your apps already run on Vercel or the AI SDK, Vercel AI Gateway is the better answer.
How do Vercel AI Gateway and Selan compare?
| Vercel AI Gateway | Selan | |
|---|---|---|
| What it is | A hosted API for apps and coding agents: text, image, video, speech and embedding models from many providers | A hosted gateway, with a CLI that launches Claude Code, Codex and OpenCode through it |
| Where it runs | Vercel's API at ai-gateway.vercel.sh; your app need not run on Vercel. Regional endpoints in the EU or the US are in beta | A hosted service, built and hosted in the EU. Nothing to run |
| Coding tools | A setup command for 29 agents, among them Claude Code, Codex, OpenCode, Cursor, GitHub Copilot CLI, Cline and Zed; Claude Desktop by hand | Claude Code, Codex, OpenCode and Claude Desktop |
| Claude for Claude Code | Vercel's own provider accounts paid from AI Gateway Credits, your own provider keys (BYOK), or each developer's own Claude subscription passed through | The Claude and ChatGPT accounts and API keys your company connects |
| A developer's credential | An AI Gateway API key, in the macOS Keychain or in ~/.claude/settings.json | One sign-in with Google or Microsoft through selan login |
| Usage | Requests, tokens, cost and every routing attempt in the AI Gateway dashboard and logs, with traces in Vercel Observability | One usage record per person, across every tool |
| Limits | Budgets per team, project, API key or member, by day, week or month in UTC or never resetting, answered with a 402 | A spend limit per person, by day, week or month in UTC, on pay-per-use keys |
| Data controls | Zero data retention, no prompt training and provider allowlists | Secret scanning replaces a secret it recognises before the request leaves |
| Cost | The provider's list price with no markup or platform fee, from prepaid credits; some add-ons cost extra | Free for up to three seats, no card |
Vercel AI Gateway gives each developer a key and bills models at the provider's list price. Selan is one sign-in, on the AI accounts your company already pays for.
When is Vercel AI Gateway the better answer?
When your own apps already run on Vercel or the AI SDK. The AI SDK has an AI Gateway provider, and a Vercel deployment can authenticate with its OIDC token instead of a stored key.
When you want many models on one bill. The gateway lists 400+ models across text, image, video, speech and embeddings. You pay the provider's list price from AI Gateway Credits, or bring your own provider keys at no markup.
When you need more coding tools. Its setup command configures 29 agents, among them Cursor, GitHub Copilot CLI, Cline, Goose, Kilo Code and Zed beside Claude Code, Codex and OpenCode.
When you want routing rules. Requests fail over across providers and fallback models, and
a virtual model in Claude Code's /model picker carries its own provider
order, fallbacks, regional inference and zero data retention. A team-wide rule can deny a
model or rewrite it to another.
If no team runs coding tools through a gateway, you do not need Selan.
How does Claude Code connect to Vercel AI Gateway?
Two ways, and the credential that reaches Anthropic decides who pays.
With an API key, a developer runs npx vercel@latest ai-gateway setup --agent claude-code.
It sets ANTHROPIC_BASE_URL to https://ai-gateway.vercel.sh/claude-code
in ~/.claude/settings.json, empties ANTHROPIC_API_KEY and turns
on gateway model discovery. On macOS it keeps the key in the Keychain and exports it as
ANTHROPIC_AUTH_TOKEN from the shell startup file. Requests are paid from the
team's AI Gateway Credits, or by your own provider key.
With a Claude subscription, the AI Gateway key moves to an x-ai-gateway-api-key
header in ANTHROPIC_CUSTOM_HEADERS. The developer logs in to Claude Code with
their own Claude account, and their plan's usage limits apply. The setup command does not
write this; it is set by hand.
The gateway's models appear in /model with a claude-code/
prefix. Claude Desktop is switched by hand, under Developer, Configure Third-Party
Inference.
How do Vercel AI Gateway budgets work?
A budget caps spend for the team, a project, one API key or one member. Every budget in scope is checked before each request, and the request is refused if any one of them is exceeded.
A member's budget covers every API key attributed to them. New keys are attributed to the member who creates them, and a default budget can cap every member, key or project that has none of its own.
Days reset at midnight UTC, weeks on Monday and months on the 1st; a budget can also
never reset. Once over, the gateway answers with an HTTP 402 of type
quota_for_entity_exceeded. Email alerts can go out at 50%, 75% and 100%.
Owners and members with the AI Gateway Budget Manager permission set team and member budgets. Anyone who can edit a key can set that key's budget.
When is Selan the better answer?
When the traffic is a team's coding tools and nobody wants to hand out keys. A developer
runs selan login with Google or Microsoft, then selan claude,
selan codex or selan opencode. Claude Desktop goes through the
same gateway with one settings file.
Claude Code, Codex, OpenCode and Claude Desktop go into one usage record per person. A limit per person applies across every pay-per-use key.
An owner connects the company's accounts and keys once, and switches on the models the team may use. No provider key reaches a laptop. With secret scanning on, a secret it recognises is replaced before the request leaves.
The trade-off is that requests travel through Selan, so if Selan is unavailable those requests fail. Your accounts can still be used directly.
Which providers serve which models is in models and providers, limits per person are on the spend page, and scanning is in the secret scanning docs.
How do you move Claude Code from Vercel AI Gateway to Selan?
Remove the marked block the setup command added to your shell startup file. Then remove
ANTHROPIC_BASE_URL, ANTHROPIC_AUTH_TOKEN,
ANTHROPIC_API_KEY, ANTHROPIC_CUSTOM_HEADERS and
CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY wherever they are set, including
the env block of ~/.claude/settings.json. The command saved a
.bak copy of each file it changed. Switch Claude Desktop's inference provider
back by hand.
Then run selan login and selan claude. Connect the company's
Anthropic accounts or keys in Selan and switch on their models. You can keep Vercel AI
Gateway for your own apps and run the team's coding tools through Selan. Keeping provider
keys off laptops is covered in
Claude Code without API keys on laptops,
and the kinds of gateway in Claude Code gateways.
Free for up to three seats, and the Free plan needs no card.
The short version
- Vercel AI Gateway is a hosted API for apps and coding agents, with 400+ models at the provider's list price and no markup.
- Claude Code reaches it with an AI Gateway API key per developer, or with the developer's own Claude subscription passed through.
- Budgets per team, project, key or member reset in UTC and refuse with a 402; they do not cover your own provider keys.
- Selan is hosted in the EU: one sign-in per person, no provider key on the laptop, and a spend limit per person.
What Vercel AI Gateway does is as Vercel's documentation (AI Gateway overview, Claude Code and
Claude Agent SDK, OpenAI Codex, Coding Agents, the vercel ai-gateway CLI
reference, API Keys, Budgets and Spend Limits, Pricing, Bring Your Own Key) and its AI Gateway
page described it, read on 12 November 2026. It changes: check the date before relying on it.
What Selan does is as its CLI and docs stated it on that date.