OpenRouter Free Model Tester

[English] | Español

A simple but powerful Node.js tool to automatically discover, filter, and test free AI models available on OpenRouter.

This project was created to answer a very practical question:

Which free models on OpenRouter actually work right now and respond correctly through the API?

OpenRouter publishes a list of models and pricing, but in practice:

• Some “free” models fail
• Some are temporarily unavailable
• Some return errors or empty responses

This app automates the full process end‑to‑end:

1. Download all models from OpenRouter
2. Filter only the 100% free ones
3. Test each free model with a real API call
4. Save a clean list of models that actually work

---

✨ Features

• Fetches the full public OpenRouter model catalog

• Filters only models with zero cost in all pricing categories

• Tests each free model with a real prompt

• Generates clean JSON outputs:

  • All models
  • Free models only
  • Working free models only

---

🧠 How the app works (Pipeline)

When you run npm start, the app executes four scripts in sequence:

1️⃣ get-all-models.js

• Calls the OpenRouter public endpoint: /api/frontend/models
• Downloads the full list of available models
• Saves the result to:

output/all-models.json

---

2️⃣ filterFreeModels.js

• Reads all-models.json

• Filters only models where all pricing fields are equal to "0":

  • prompt
  • completion
  • image
  • image_output
  • internal_reasoning

• Extracts and cleans useful fields:

  • name
  • model_variant_slug
  • description
  • context length
  • modalities

• Saves the result to:

output/free-models.json

---

3️⃣ test-models.js

• Reads free-models.json

• For each free model:

  • Sends a test prompt to /api/v1/chat/completions
  • Uses your OpenRouter API key
  • Checks if the model returns a valid answer

Test prompt used:

"Hello, how are you? Answer in less than 10 words."

• Successful models are marked as working

• Failed models are logged with error details

• Saves only the working models to:

output/models-working.json

---

4️⃣ start-check-models.js

This is the orchestrator script.

It runs, in order:

1. get-all-models.js
2. filterFreeModels.js
3. test-models.js

And prints a final summary when finished.

---

📦 Installation

Clone the repository and install dependencies:

npm install

---

🔑 Environment configuration

You must provide your OpenRouter API key using a .env file.

Create a file named .env in the root of the project:

OPENROUTER_API_KEY=sk-or-v1-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

⚠️ Important:

• Do NOT use quotes
• Do NOT add semicolons

Correct:

OPENROUTER_API_KEY=sk-or-v1-abc123

Incorrect:

OPENROUTER_API_KEY="sk-or-v1-abc123";

---

▶️ How to run

Simply execute:

npm start

This runs:

node start-check-models.js

The full pipeline will execute automatically.

---

📁 Output files

After running, you will find:

output/
 ├─ all-models.json        # Full OpenRouter catalog
 ├─ free-models.json       # Only 100% free models
 └─ models-working.json   # Free models that actually work

models-working.json is the final and most useful result.

---

🛠️ Requirements

• Node.js 18+ (or any version with native fetch support)
• An OpenRouter account and API key

---

⚡ Notes

• Requests are executed in parallel. If OpenRouter rate‑limits you, consider adding concurrency control.
• Some models may temporarily fail depending on availability.
• This tool uses only public and official OpenRouter endpoints.

---

🎯 Motivation

This project was created to:

• Discover which free models are truly usable
• Avoid wasting time manually testing models
• Build an automated benchmark of free AI availability
• Help developers quickly choose working free models

---

📜 License

MIT License

---

🙌 Author

Created by Daniel Rincon — Founder Nahoral Labs