llm-poe

Plugin for LLM adding support for Poe API models. This plugin dynamically fetches all available models from Poe's API, ensuring you always have access to the latest models.

✅ Full Multi-Modal Support: Works with text, image generation, video generation, and audio/TTS models with optimized handling for each content type.

✅ 94% Test Coverage: Comprehensive test suite with 136 automated tests ensuring reliability.

Installation

Install this plugin in the same environment as LLM:

llm install llm-poe

Or install from source for development:

llm install -e .

Important: Use llm install -e . (not pip install -e .) for local development to ensure proper plugin discovery.

Configuration

You need to set your Poe API key. Get your API key from https://poe.com/api_key

Set the key using:

llm keys set poe

Or set the POE_API_KEY environment variable:

export POE_API_KEY="your-api-key-here"

Usage

List available models:

llm models list | grep poe/

The plugin automatically fetches all available models from Poe's API. The model list is cached for 1 hour to improve performance. Over 240+ models are currently available including the latest AI models from OpenAI, Anthropic, Google, Meta, and more.

Run prompts with different model types:

# Text generation
llm -m poe/gpt_4o "Hello, how are you?"

# Image generation
llm -m poe/flux_pro_1_1_ultra "Generate a cat image" --no-stream

# Video generation  
llm -m poe/sora "Create a short video of a sunset" --no-stream

# Audio/TTS
llm -m poe/elevenlabs_v3 "Say hello world" --no-stream

Available Models

The plugin dynamically fetches all models available through Poe's API, including:

Latest Text/Code Models:

• GPT-4.1, O3-Pro, O4-Mini, O3-Mini
• Claude-Sonnet-4, Claude-Opus-4.1
• Gemini-2.5-Pro, Gemini-2.5-Flash
• DeepSeek-R1, DeepSeek-V3
• Grok-4, Grok-3
• Llama-4-Scout, Llama-4-Maverick
• Qwen-3-235B, Mistral-Small-3.2

Image Generation Models:

• Imagen-4-Ultra, Flux-Pro-1.1
• DALL-E-3, Ideogram-v3
• Phoenix-1.0, Recraft-v3
• Flux-Kontext-Max

Video Generation Models:

• Veo-3, Sora, Runway-Gen-4-Turbo
• Kling-2.1, Hailuo-02
• Dream-Machine, Pika

Audio/TTS Models:

• ElevenLabs, Cartesia
• PlayAI-TTS, Orpheus-TTS
• Lyria, Hailuo-Speech

Search-Enhanced Models:

• GPT-4o-Search, Claude-Sonnet-4-Search
• Perplexity-Sonar-Pro, Web-Search

And 200+ more models! Use llm models list | grep poe/ to see the complete current list.

Options

Text Models:

llm -m poe/claude_sonnet_4 "Tell me a story" -o temperature 0.7 -o max_tokens 500

Image Models:

llm -m poe/flux_pro_1_1_ultra "Generate landscape" -o size 1024x1024 -o quality standard --no-stream

Video Models:

llm -m poe/sora "Create a nature scene" -o duration 10 -o aspect_ratio 16:9 --no-stream

Audio Models:

llm -m poe/elevenlabs_v3 "Read this text" -o voice alloy -o speed 1.0 --no-stream

Development

To set up for development:

git clone https://github.com/mrf/llm-poe
cd llm-poe
llm install -e .

Important: Use llm install -e . (not pip install -e .) for local development to ensure proper plugin discovery.

Testing

This plugin has a comprehensive test suite with 94% code coverage.

Running Tests

# Install with test dependencies
pip install -e ".[test]"

# Run all tests
pytest

# Run with coverage report
pytest --cov=llm_poe --cov-report=term-missing --cov-report=html

# View HTML coverage report
open htmlcov/index.html  # macOS
xdg-open htmlcov/index.html  # Linux

Test Coverage

• 136 automated tests covering all functionality
• 94% code coverage across the entire plugin
• Multi-platform testing (Ubuntu, macOS, Windows)
• Python 3.9-3.12 compatibility verified

Test categories:

• Unit tests for model registration, API keys, and type detection
• Integration tests for all model types (text, image, video, audio)
• Error handling and edge case tests
• Performance and caching tests

See tests/README.md for detailed testing documentation.

License

Apache-2.0

---

This plugin was created with assistance from Claude.