Kimi CLI with AI Gateway

Kimi CLI is Moonshot AI's terminal coding agent. It reads a TOML config at $KIMI_SHARE_DIR/config.toml (default ~/.kimi), which you can point at AI Gateway for unified model access and spend monitoring.

Run the Vercel CLI setup command for Kimi CLI:

npx vercel ai-gateway setup --agent kimi

The command writes a gateway provider and a [models] shortlist into $KIMI_SHARE_DIR/config.toml, with Moonshot's newest models first and the gateway's most-used models after them. It sets default_model to prefer moonshotai/kimi-k3. The file is created with 0600 permissions because Kimi requires a literal API key.

If you can't use the Vercel CLI, configure Kimi CLI manually:

1. Create an API key

   Go to the AI Gateway section in the Vercel dashboard sidebar and click API keys to create a new API key.

2. Add the gateway provider

   Add the provider and a [models] shortlist to ~/.kimi/config.toml:

   ~/.kimi/config.toml

   default_model = "moonshotai/kimi-k3"
    
   [providers.vercel-ai-gateway]
   name = "Vercel AI Gateway"
   base_url = "https://ai-gateway.vercel.sh/coding-agent/v1"
   api_key = "your-ai-gateway-api-key"
    
   [models."moonshotai/kimi-k3"]
   provider = "vercel-ai-gateway"
   model = "moonshotai/kimi-k3"
   max_context_size = 1000000
   display_name = "Kimi K3"
    
   [models."openai/gpt-6-astra"]
   provider = "vercel-ai-gateway"
   model = "openai/gpt-6-astra"
   max_context_size = 1050000
   display_name = "GPT-6 Astra"

   Kimi providers need a literal api_key, so the key lands in this file. Create it with mode 0600 so only you can read it:

   Terminal

   chmod 0600 ~/.kimi/config.toml

   default_model must name a key declared under [models].

3. Start Kimi CLI

   Terminal

   kimi

   Switch models in-session. Any gateway model ID works by adding a matching [models."<gateway-model-id>"] entry. Quote the key because TOML bare keys can't contain /.

View your usage, spend, and request activity in the AI Gateway section in the Vercel dashboard sidebar. See the observability documentation for more details.

• Browse the models catalog for available models
• Configure spend limits on your API key