Local Multi-LLM Testing & Performance Tracker

This n8n workflow enables developers, researchers, and marketers to efficiently benchmark multiple local Large Language Models (LLMs) running via LM Studio. It dynamically fetches active models, applies configurable parameters like temperature and top P, sends test prompts to each model, captures responses, and computes key performance metrics including response time, word count, and readability scores (e.g., Flesch-Kincaid ranges from very easy to standard readability).

Results are automatically logged into a Google Sheets spreadsheet, allowing for seamless comparison across models and settings. This eliminates manual testing, providing structured data for analysis, visualization, and decision-making on optimal models for tasks like ad copy generation or audience targeting.

Ideal for marketing teams optimizing AI-generated content for ads and conversions, or AI enthusiasts iterating on local setups. Benefits include time savings on repetitive tests, objective performance insights, and scalability for ongoing model evaluations. Minimal setup required: connect to LM Studio IP, create a Google Sheet, and run.

Use cases span AI research, prompt engineering for marketing automation, and performance tracking in resource-constrained environments without cloud costs.