Skip to content

Vercel AI gateway

The Vercel AI Gateway provides a unified API to access hundreds of models through a single endpoint.

PropertyValue
Providervercel-ai-gateway
AuthAI_GATEWAY_API_KEY
APIAnthropic Messages compatible
Model catalogAuto-discovered via /v1/models
  1. Set the API key

    Run onboarding and choose the AI Gateway auth option:

    Terminal window
    openclaw onboard --auth-choice ai-gateway-api-key
  2. Set a default model

    Add the model to your OpenClaw config:

    {
    agents: {
    defaults: {
    model: { primary: "vercel-ai-gateway/anthropic/claude-opus-4.6" },
    },
    },
    }
  3. Verify the model is available

    Terminal window
    openclaw models list --provider vercel-ai-gateway

For scripted or CI setups, pass all values on the command line:

Terminal window
openclaw onboard --non-interactive \
--mode local \
--auth-choice ai-gateway-api-key \
--ai-gateway-api-key "$AI_GATEWAY_API_KEY"

OpenClaw accepts Vercel Claude shorthand model refs and normalizes them at runtime:

Shorthand inputNormalized model ref
vercel-ai-gateway/claude-opus-4.6vercel-ai-gateway/anthropic/claude-opus-4.6
vercel-ai-gateway/opus-4.6vercel-ai-gateway/anthropic/claude-opus-4-6
Environment variable for daemon processes

If the OpenClaw Gateway runs as a daemon (launchd/systemd), make sure AI_GATEWAY_API_KEY is available to that process.

Provider routing

Vercel AI Gateway routes requests to the upstream provider based on the model ref prefix. For example, vercel-ai-gateway/anthropic/claude-opus-4.6 routes through Anthropic, while vercel-ai-gateway/openai/gpt-5.5 routes through OpenAI and vercel-ai-gateway/moonshotai/kimi-k2.6 routes through MoonshotAI. Your single AI_GATEWAY_API_KEY handles authentication for all upstream providers.

Thinking levels

/think options follow trusted upstream model prefixes when OpenClaw knows the upstream provider contract. vercel-ai-gateway/anthropic/... uses the Claude thinking profile, including adaptive defaults for Claude 4.6 models. vercel-ai-gateway/openai/gpt-5.4, gpt-5.5, and Codex-style refs expose /think xhigh just like the direct OpenAI/OpenAI Codex providers. Other namespaced refs keep the normal reasoning levels unless their catalog metadata declares more.

Model selection

Choosing providers, model refs, and failover behavior.

Troubleshooting

General troubleshooting and FAQ.