Prerequisites
- Kimi Code installed:
- Active Orq.ai account with AI Gateway access
- Orq.ai API key
Setup
Kimi Code is OpenAI-compatible. Edit~/.kimi-code/config.toml to add an Orq.ai provider pointed at the router, and register each model as its own [models."..."] block:
<ORQ_API_KEY> with an API key. If ~/.kimi-code/config.toml does not exist yet, create it. Each model value is an identifier from the AI Gateway catalog, so add a [models."..."] block for every model you want. default_model sets the one loaded at startup; switch between registered models at any time with /model inside Kimi Code.
Launch Kimi Code:
/provider inside Kimi Code.
Verification
Once configured, send a prompt in Kimi Code. The response appears in the terminal and the trace appears in Orq.ai Traces with the model identifier you selected.Connect via MCP
Kimi Code also acts as an MCP client, so the Orq MCP server can run alongside the gateway routing above. With both connected, Kimi Code routes model calls through the AI Gateway and manages Agents, experiments, Datasets, and Traces through the Orq MCP tools. Add the Orq MCP server to~/.kimi-code/mcp.json (or the project-local .kimi-code/mcp.json):
YOUR_ORQ_API_KEY with an API key. Manage servers from the terminal with the kimi mcp command group. See the MCP Quickstart for the full tool reference.
Orq.ai API keys are scoped to a single project, so the Orq MCP server only exposes the Agents, experiments, Datasets, and Traces within that project. Use a key from the project whose entities Kimi Code should manage.