Vercel AI Gateway Provider Setup and Model Selection
Learn how to configure the Vercel AI Gateway provider in OpenClaw, including authentication, model discovery, and usage examples. This guide is for developers integrating hundreds of AI models through a single endpoint.
Read this when
- You want to use Vercel AI Gateway with OpenClaw
- You need the API key env var or CLI auth choice
The Vercel AI Gateway offers a single endpoint that exposes a unified API for reaching hundreds of models.
| Property | Value |
|---|---|
| Provider | vercel-ai-gateway |
| Package | @openclaw/vercel-ai-gateway-provider |
| Auth | AI_GATEWAY_API_KEY |
| API | Anthropic Messages compatible |
| Base URL | https://ai-gateway.vercel.sh |
| Model catalog | Auto-discovered via /v1/models |
Tip
Because OpenClaw automatically discovers the Gateway
/v1/modelscatalog, both the/models vercel-ai-gatewaychat command andopenclaw models list --provider vercel-ai-gatewaycarry current model refs likevercel-ai-gateway/openai/gpt-5.5andvercel-ai-gateway/moonshotai/kimi-k2.6.
Getting started
Install the plugin
openclaw plugins install @openclaw/vercel-ai-gateway-provider
Set the API key
openclaw onboard --auth-choice ai-gateway-api-key
Set a default model
{
agents: {
defaults: {
model: { primary: "vercel-ai-gateway/anthropic/claude-opus-4.6" },
},
},
}
Verify the model is available
openclaw models list --provider vercel-ai-gateway
Non-interactive example
openclaw onboard --non-interactive --accept-risk --skip-health \
--mode local \
--auth-choice ai-gateway-api-key \
--ai-gateway-api-key "$AI_GATEWAY_API_KEY"
Model ID shorthand
At runtime, OpenClaw normalizes Claude shorthand model refs:
| Shorthand input | Normalized model ref |
|---|---|
vercel-ai-gateway/claude-opus-4.6 | vercel-ai-gateway/anthropic/claude-opus-4.6 |
vercel-ai-gateway/opus-4.6 | vercel-ai-gateway/anthropic/claude-opus-4-6 |
Tip
Either form works in your configuration; the canonical
anthropic/...ref gets resolved automatically by OpenClaw.
Advanced configuration
Environment variable for daemon processes
When the OpenClaw Gateway operates as a daemon (launchd/systemd), ensure
AI_GATEWAY_API_KEY is accessible to that process.
Warning
A key set only in an interactive shell stays hidden from a launchd/systemd daemon unless that environment is explicitly imported. To guarantee the gateway process can access it, place the key in
~/.openclaw/.envor set it viaenv.shellEnv.
Provider routing
Each request through Vercel AI Gateway is forwarded to the upstream provider indicated by the model ref prefix. As an example, vercel-ai-gateway/anthropic/claude-opus-4.6
goes to Anthropic, vercel-ai-gateway/openai/gpt-5.5 goes to
OpenAI, and vercel-ai-gateway/moonshotai/kimi-k2.6 goes to
MoonshotAI. A single AI_GATEWAY_API_KEY handles authentication for every upstream provider.
Thinking levels
When OpenClaw recognizes the upstream model prefix, /think options follow it.
vercel-ai-gateway/anthropic/... applies the Claude thinking profile,
including the adaptive default for Claude 4.6 models. Trusted
vercel-ai-gateway/openai/... refs (gpt-5.2 and later, plus Codex
variants back to gpt-5.1-codex) expose /think xhigh. Other namespaced
refs retain the standard reasoning levels unless their catalog metadata
specifies more.
Related
-
Model selection, Picking providers, model refs, and failover behavior.
-
Troubleshooting, FAQ and general troubleshooting.