Use LithosAI with Pi

Before you begin

Create a key on the API Keys page, confirm the model ID on the Models page, and export the key in the shell where you start Pi.

export LITHOSAI_API_KEY="your-key"

Pi reaches LithosAI directly through Chat Completions. Install Pi, then add a provider to ~/.pi/agent/models.json.

npm install -g --ignore-scripts @earendil-works/pi-coding-agent

~/.pi/agent/models.json

{
  "providers": {
    "lithosai": {
      "baseUrl": "https://api.lithosai.cloud/v1",
      "api": "openai-completions",
      "apiKey": "$LITHOSAI_API_KEY",
      "models": [
        {
          "id": "moonshotai/Kimi-K3",
          "name": "Kimi K3 on LithosAI"
        },
        {
          "id": "moonshotai/Kimi-K3-fast",
          "name": "Kimi K3 fast on LithosAI"
        }
      ]
    }
  }
}

Start Pi with a LithosAI model. You can also run /model inside Pi to switch between the configured models.

pi --provider lithosai --model moonshotai/Kimi-K3

Pi uses default context and output limits when they are omitted from models.json. Check the Models console page and add contextWindow and maxTokens to each entry when you need Pi’s context meter and compaction behavior to match the selected model exactly.

This connection uses Pi’s OpenAI Chat Completions implementation. Provider-specific reasoning controls require additional Pi compatibility settings and are not enabled by this minimal configuration.

Troubleshooting

Symptom	Cause
The LithosAI models do not appear in `/model`	`LITHOSAI_API_KEY` is unset, or `apiKey` does not include the `$` environment-variable prefix.
The context meter is inaccurate	Add the model’s `contextWindow` and `maxTokens` values from the LithosAI console.
`invalid 'parameters' schema`	An enabled tool or MCP server declares parameters that are not a JSON Schema object. Disable it or correct its schema.

Issues common to every agent are listed under Common issues.

Getting Started

Inference APIs

Account

Use LithosAI with Pi

Before you begin

Troubleshooting

​Before you begin

​Troubleshooting

Before you begin

Troubleshooting