Skip to main content

Guide for Interface Configuration

This guide walks you through setting up Poe API with popular AI coding tools. With Poe's OpenAI-compatible API, you can access hundreds of AI models including GPT-4o, Claude Sonnet 4, Gemini 2.5 Pro, and more through your existing coding workflow.

Prerequisites​

Before setting up any interface, you'll need:

  1. Poe API Key: Get your API key from poe.com/api_key
  2. Poe Subscription: An active Poe subscription to access the models
  3. Base URL: https://api.poe.com/v1

Available Models​

Popular models you can use include:

  • GPT-4o - OpenAI's latest model
  • Claude-Sonnet-4 - Anthropic's most capable model
  • Gemini-2.5-Pro - Google's flagship model
  • Llama-3.1-405B - Meta's largest open-source model
  • Grok-4 - xAI's latest model

For a complete list, visit the models endpoint or check available models in the Poe web interface.

Cursor Setup​

Cursor is an AI-powered code editor built for pair-programming with AI.

Step-by-Step Configuration​

  1. Open Cursor Settings

    • Press Cmd/Ctrl + , to open settings
    • Or go to Cursor → Preferences (Mac) / File → Preferences (Windows/Linux)

  2. Navigate to Models

    • Click on Models in the left sidebar
    • Look for the Chat or AI Models section

  3. Add Custom Model Provider

    • Click Add Model or Configure Custom Provider
    • Select OpenAI Compatible or Custom OpenAI

  4. Configure Poe API Settings

    Provider Name: Poe
    Base URL: https://api.poe.com/v1
    API Key: [Your Poe API Key from poe.com/api_key]

  5. Select a Model

    • Choose from available models like:
      • Claude-Sonnet-4
      • GPT-4o
      • Gemini-2.5-Pro

  6. Test the Configuration

    • Open a new chat with Cmd/Ctrl + L
    • Send a test message to verify the setup

Usage Tips for Cursor​

  • Use Cmd/Ctrl + K for inline code generation
  • Use Cmd/Ctrl + L for chat-based assistance
  • The AI can see your entire codebase context automatically

Cline Setup​

Cline is an autonomous coding agent that works as a VS Code extension.

Step-by-Step Configuration​

  1. Install Cline Extension

    • Open VS Code
    • Go to Extensions (Cmd/Ctrl + Shift + X)
    • Search for "Cline" and install it

  2. Open Cline Settings

    • Click the Cline icon in the Activity Bar
    • Or press Cmd/Ctrl + Shift + P and search "Cline: Open"

  3. Configure API Settings

    • In the Cline panel, click the settings/gear icon
    • Select Custom OpenAI Compatible as the provider

  4. Enter Poe Configuration

    API Provider: OpenAI Compatible
    Base URL: https://api.poe.com/v1
    API Key: [Your Poe API Key]
    Model: Claude-Sonnet-4

  5. Verify Setup

    • Start a new conversation with Cline
    • Give it a simple coding task to test

Usage Tips for Cline​

  • Cline can read, write, and execute code autonomously
  • It works best with specific, well-defined tasks
  • Always review code changes before accepting them

Roo Code Setup​

Roo Code is an AI coding assistant that integrates with various development environments.

Step-by-Step Configuration​

  1. Access Settings

    • Open Roo Code settings/preferences
    • Look for AI Provider or LLM Configuration section

  2. Choose Custom Provider

    • Select Custom OpenAI or OpenAI Compatible option
    • Choose Add Custom Provider if available

  3. Configure Poe Settings

    Provider: Custom OpenAI
    API Endpoint: https://api.poe.com/v1
    API Key: [Your Poe API Key]
    Model: GPT-4o

  4. Advanced Settings (if available)

    Temperature: 0.7
    Max Tokens: 4000
    Stream: Enabled

  5. Test Connection

    • Use the test/verify button if available
    • Or start a new conversation to confirm setup

Usage Tips for Roo Code​

  • Experiment with different models for different tasks
  • Use streaming for real-time responses
  • Adjust temperature based on creativity needs (lower for code, higher for ideation)

LLM (Command Line Tool) Setup​

LLM is a command-line tool for working with Large Language Models. It provides a simple way to prompt various AI models directly from your terminal.

Step-by-Step Configuration​

  1. Install LLM

    Choose one of the following installation methods:

    # Using pip
    pip install llm

    # Using pipx (recommended for CLI tools)
    pipx install llm

    # Using uv
    uv tool install llm

    # Using Homebrew (macOS/Linux)
    brew install llm

  2. Configure Poe API Key

    Set your Poe API key using LLM's key management:

    llm keys set poe
    # Enter your Poe API key when prompted

    Alternatively, use an environment variable:

    export POE_API_KEY='your_poe_api_key_here'

  3. Test Basic Setup

    Try a simple prompt to verify the setup:

    # Using stored key
    llm "Hello, world!" --key poe --api-base https://api.poe.com/v1

    # Using environment variable
    llm "Hello, world!" --key $POE_API_KEY --api-base https://api.poe.com/v1

  4. Create Model Aliases

    For easier access, create aliases for your favorite Poe models:

    # Set Claude Sonnet 4 as default
    llm models default claude-sonnet-4 --api-base https://api.poe.com/v1 --key poe

    # Or create specific aliases
    llm alias claude "Claude-Sonnet-4" --api-base https://api.poe.com/v1 --key poe
    llm alias gpt4 "GPT-4o" --api-base https://api.poe.com/v1 --key poe

  5. Configuration File Setup

    For persistent configuration, you can set up a configuration file. The config location varies by OS:

    • macOS: ~/Library/Application Support/io.datasette.llm/
    • Linux: ~/.config/io.datasette.llm/

    Create or edit the configuration to include Poe settings.

Usage Examples​

Basic prompting:

# Simple prompt with specific model
llm "Explain quantum computing" -m Claude-Sonnet-4 --api-base https://api.poe.com/v1 --key poe

# Using alias (if configured)
llm "Write a Python function to calculate fibonacci" -m claude

Interactive chat:

# Start an interactive session
llm chat -m GPT-4o --api-base https://api.poe.com/v1 --key poe

File input:

# Process a file
cat mycode.py | llm "Review this code for bugs" -m Claude-Sonnet-4 --api-base https://api.poe.com/v1 --key poe

# Save conversation to file
llm "Explain machine learning" -m Gemini-2.5-Pro --api-base https://api.poe.com/v1 --key poe > explanation.txt

Using templates:

# Create a template for code review
llm template code-review "Review this code and suggest improvements: {code}"

# Use the template
llm -t code-review -v code "$(cat script.py)" -m Claude-Sonnet-4 --api-base https://api.poe.com/v1 --key poe

Environment Setup​

For convenience, you can set these environment variables in your shell profile:

# Add to ~/.bashrc, ~/.zshrc, or equivalent
export POE_API_KEY='your_poe_api_key_here'
export LLM_API_BASE='https://api.poe.com/v1'

# Create aliases for common commands
alias llm-poe='llm --api-base $LLM_API_BASE --key $POE_API_KEY'
alias llm-claude='llm -m Claude-Sonnet-4 --api-base $LLM_API_BASE --key $POE_API_KEY'
alias llm-gpt='llm -m GPT-4o --api-base $LLM_API_BASE --key $POE_API_KEY'

Usage Tips for LLM​

  • Logging: LLM automatically logs conversations to SQLite. Use llm logs off to disable or llm logs on to enable
  • Multiple Models: Easily switch between Poe models using the -m flag
  • Streaming: LLM supports streaming responses for real-time output
  • Plugins: Extend functionality with LLM plugins for specialized tasks
  • Templates: Create reusable prompt templates for common workflows
  • Pipes: LLM works great with Unix pipes for processing files and command output

Advanced Configuration​

Custom model configuration:

# List available models from Poe
curl -H "Authorization: Bearer $POE_API_KEY" https://api.poe.com/v1/models

# Use any model from the list
llm "Your prompt here" -m "Llama-3.1-405B" --api-base https://api.poe.com/v1 --key poe

Batch processing:

# Process multiple prompts from a file
while IFS= read -r prompt; do
  echo "Prompt: $prompt"
  llm "$prompt" -m Claude-Sonnet-4 --api-base https://api.poe.com/v1 --key poe
  echo "---"
done < prompts.txt

Troubleshooting​

Common Issues​

"Invalid API Key" Error

  • Verify your API key from poe.com/api_key
  • Ensure there are no extra spaces or characters
  • Check that your Poe subscription is active

"Model Not Found" Error

  • Verify the model name is spelled correctly
  • Use exact model names like Claude-Sonnet-4 (case-sensitive)
  • Check if the model is available with your subscription

Connection Timeout

  • Verify the base URL: https://api.poe.com/v1
  • Check your internet connection
  • Try a different model if one seems unresponsive

Rate Limiting

  • Poe has a 500 requests per minute limit
  • Wait a moment before retrying
  • Consider upgrading your subscription for higher limits

Getting Help​

Benefits of Using Poe API​

  • Single API Key: Access hundreds of models with one key
  • Cost Effective: Use existing Poe subscription points
  • Model Variety: Switch between different providers and models
  • Up-to-date Models: Access the latest models as they're released
  • Reliable Infrastructure: Built for scale and performance