Pi OpenRouter Extension


OpenRouter provider extension for Pi Coding Agent.

Features

- 🚀 Access 20+ LLM models through OpenRouter's unified API
- 🔄 OpenAI-compatible streaming
- 🤖 Supports Claude, GPT-4, Gemini, Llama, Mistral, DeepSeek, Perplexity
- 🆓 Free models available (Gemini 2.0 Flash)
- 📦 One-command installation via npm
- ⚡ Auto-installs to ~/.pi/agent/extensions/

Installation

$3

``bash
npm install -g @verioussmith/pi-openrouter
`

That's it! The extension automatically installs to ~/.pi/agent/extensions/openrouter.ts

$3

See INSTALL.md for manual installation options.

Configuration

Set your OpenRouter API key in ~/.zshrc or environment:

`bash
export OPENROUTER_API_KEY="sk-or-v1-YOUR_KEY_HERE"
`

Get your key at: https://openrouter.ai/settings/keys

Usage

`bash


Start pi

pi

List available models

/model

Select an OpenRouter model

/model openrouter/google/gemini-2.0-flash-exp:free

Or any other model

/model openrouter/anthropic/claude-3.5-sonnet
/model openrouter/openai/gpt-4o
/model openrouter/deepseek/deepseek-r1
`

Available Models

$3

- google/gemini-2.0-flash-exp:free - Gemini 2.0 Flash Experimental
- google/gemini-2.0-flash-thinking-exp:free - Gemini 2.0 Flash with reasoning

$3

- anthropic/claude-3.5-sonnet - Claude 3.5 Sonnet
- anthropic/claude-3-opus - Claude 3 Opus
- anthropic/claude-3-haiku - Claude 3 Haiku

$3

- openai/gpt-4o - GPT-4o
- openai/gpt-4o-mini - GPT-4o Mini
- openai/o1 - O1 (reasoning model)
- openai/o1-mini - O1 Mini

$3

- google/gemini-pro-1.5 - Gemini Pro 1.5

$3

- meta-llama/llama-3.3-70b-instruct - Llama 3.3 70B
- meta-llama/llama-3.1-405b-instruct - Llama 3.1 405B

$3

- mistralai/mistral-large - Mistral Large
- mistralai/mistral-small - Mistral Small

$3

- deepseek/deepseek-chat - DeepSeek Chat
- deepseek/deepseek-r1 - DeepSeek R1 (reasoning)

$3

- perplexity/sonar-pro - Sonar Pro

See all models: https://openrouter.ai/models

Adding More Models

Edit openrouter.ts and add to the models array:

`typescript
{
id: "provider/model-name",
name: "Display Name (OpenRouter)",
reasoning: false, // true for reasoning models
input: ["text", "image"], // or just ["text"]
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
contextWindow: 128000,
maxTokens: 4096,
}
`

Costs are in $/million tokens. Check https://openrouter.ai/models for pricing.

Development

`bash


Clone

git clone https://github.com/verioussmith/pi-openrouter-extension.git
cd pi-openrouter-extension

Install in Pi (development mode)

ln -s $(pwd)/openrouter.ts ~/.pi/agent/extensions/openrouter.ts

Test

pi
/model openrouter/google/gemini-2.0-flash-exp:free
`

$3

`bash


Update version in package.json

npm version patch # or minor, major

Commit and push

git add . && git commit -m "chore: bump version" && git push

Publish to npm (requires OTP)

npm publish --access public --otp=YOUR_OTP
``

Package Links

- npm: https://www.npmjs.com/package/@verioussmith/pi-openrouter
- GitHub: https://github.com/verioussmith/pi-openrouter-extension
- Issues: https://github.com/verioussmith/pi-openrouter-extension/issues

References

- Pi Coding Agent
- Pi Custom Provider Docs
- OpenRouter API
- OpenRouter Models

Contributing

Contributions welcome! Please:
1. Fork the repo
2. Create a feature branch
3. Submit a pull request

Author

Verious Smith (@verioussmith)
- Website: https://philoveracity.com
- Email: verious@philoveracity.com

License

MIT - see LICENSE file for details