Lightspeed - GPT4 @133 tokens/s

github.com

Multithreading for OpenAI calls with robust error handling

Biggest problems with OpenAI GPT-4 API: 1. API errors (servers overloaded, request failed, no engine found, etc.). Don't let one API call failure ruin your entire program! 2. Slow! Lightspeed is an open source free solution available as a CLI tool and Colab.