Compare AI models in one place and score how easy their answers are to read. Send a prompt once, get results from many models, and log the scores to Google Sheets. Useful for marketing and content teams that want clear, simple text at scale.

A chat message starts the run. The flow calls your local LM server to fetch active model IDs, then splits the list to run each model separately. It records the start time, adds a system prompt for short fifth grade text, and calls each model through an OpenAI compatible endpoint. A chain then analyzes the reply for readability, word count, sentence count, and average sentence length. The flow captures end time, calculates total duration, prepares clean fields, and writes everything to Google Sheets.

You need LM Studio running with models loaded and the base URL set to your local IP. Connect OpenAI and Google Sheets credentials in n8n. Expect faster model checks, less copy and paste, and a repeatable scorecard your team can share. Use it to pick models for landing pages, support replies, and training content. Many teams cut review time from hours to minutes and compare new models in one pass.

• Chat message trigger starts the evaluation on demand
• HTTP request fetches active model IDs from your LM server
• Split out runs each model separately for clean comparisons
• System prompt enforces concise fifth grade tone for fair tests
• Model node calls an OpenAI compatible endpoint with dynamic model IDs
• Analysis chain scores readability and basic text structure
• Date and time nodes capture start, end, and total duration
• Code step aggregates metrics and prepares clean fields
• Google Sheets node appends results to a mapped spreadsheet

• Reduce manual testing time from 2 hours to 5 minutes
• Automate 90 percent of repetitive scoring work
• Handle many models in one run with no extra clicks
• Improve data quality by logging straight to Google Sheets
• Keep tests consistent with one shared system prompt
• See response speed and readability in the same report

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Google Sheets and OpenAI. See the Tools Required section above for links to create accounts with these services.
3. Install and start LM Studio on your machine. Load the models you want to evaluate so they are available to the local server.
4. Open the Get Models node and update the Base URL to your LM server IP and port. Execute the node to confirm it returns a list of model IDs.
5. Leave the Extract Model IDs step as is. It splits the returned list so each model runs in its own path.
6. Open Add System Prompt and set clear rules for short, simple, fifth grade text. This keeps tests fair across models.
7. Open the Run Model with Dynamic Inputs node. Confirm the baseURL points to your LM server, model uses the dynamic expression, and settings like temperature, topP, and presence penalty match your test plan.
8. Set up the OpenAI credential: double click the model node, choose Create new credential in the Credential to connect with dropdown, then follow the on screen instructions. When using LM Studio, a placeholder API key is often acceptable.
9. Create a Google Sheet with headers for Prompt, Time Sent, Time Received, Total Time Spent, Model, Response, Readability Score, Average Word Length, Word Count, Sentence Count, and Average Sentence Length.
10. Open the Google Sheets node, create a new Google Sheets OAuth credential, select your spreadsheet and sheet, then map each field to the correct column.
11. Open the Chat window in n8n and send a test prompt. Check that Google Sheets receives a new row for each model.
12. If no models appear, verify your LM server IP and port, confirm models are loaded, and rerun Get Models. If results seem mixed, clear the previous chat session before testing again.
13. Optional: If you prefer manual review, remove or disable the Google Sheets node and inspect results in the execution data.