Pi

Pi is a minimal, extensible coding agent for the terminal. Pi reads extra providers from a models.json file, and this guide adds Entrim to it.

Prerequisites

To follow this guide, you need:

• An Entrim account with Model API credits
• An Entrim API key
• Node.js with npm, to install Pi

---

Set up Pi

1. Install Pi

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

This installs the pi command. The Pi docs list other ways to install it.

2. Add Entrim to models.json

Create ~/.pi/agent/models.json with the entrim provider below. If the file already exists, add entrim to its providers object.

~/.pi/agent/models.json

{
  "providers": {
    "entrim": {
      "baseUrl": "https://api.entrim.ai/v1",
      "api": "openai-completions",
      "apiKey": "$ENTRIM_API_KEY",
      "models": [
        {
          "id": "zai-org/GLM-5.3-Flash",
          "name": "GLM-5.3-Flash",
          "reasoning": true,
          "input": ["text", "image"],
          "contextWindow": 1048576,
          "maxTokens": 131072
        }
      ]
    }
  }
}

"apiKey": "$ENTRIM_API_KEY" tells Pi to read the key from the ENTRIM_API_KEY environment variable each time it sends a request, so the key itself stays out of the file.

Field	Meaning
api	openai-completions makes Pi use the chat completions API
reasoning	The model reasons before it answers, so Pi offers thinking levels for it
input	The model accepts text and images
contextWindow, maxTokens	The model's limits, from max_input_tokens and max_output_tokens in GET /v1/models

3. Set your API key

Export the key in your shell profile, such as ~/.zshrc or ~/.bashrc, so it's set in every terminal:

export ENTRIM_API_KEY="<your-entrim-api-key>"

Open a new terminal, or run source ~/.zshrc, before you start Pi.

4. Select the model

Start pi, run /model and choose GLM-5.3-Flash from the entrim provider. Press Ctrl+S on it to make it the default for new sessions. Use /thinking to change how much the model reasons.

You can also pick the model when you start Pi:

pi --model entrim/zai-org/GLM-5.3-Flash

---

Check that it works

List the Entrim models Pi found, then send a one-off prompt:

pi --list-models entrim
pi -p --model entrim/zai-org/GLM-5.3-Flash "Reply with the word pong"

The first command shows the model with its context size and capabilities. The second prints the model's reply and exits.

Add more models

To use another Entrim model, add another object to the models array. Take its id from GET /v1/models, set contextWindow and maxTokens from max_input_tokens and max_output_tokens, and set reasoning and input to match what the model supports. Its page in the model library lists its capabilities. Pi reloads models.json every time you open /model.

Troubleshooting

Problem	Fix
The Entrim model doesn't appear in /model	Pi hides models it has no key for. Check that ENTRIM_API_KEY is exported in the terminal you started Pi from.
401 errors	ENTRIM_API_KEY holds an invalid key. Copy it again from the dashboard.
400 "Invalid model name"	The model id doesn't match an Entrim model ID. Compare it with GET /v1/models, including upper and lower case.