๐ Getting Started with AIQ-X
๐ Step 1: Understanding Test Packs
Test packs contain questions that evaluate specific AI capabilities across different domains like logic, math, creativity, and ethics. Each pack has three tiers: Basic (quick), Advanced (thorough), and Expert (comprehensive). Start with importing a test pack from the /Test-Packs Library.
๐ค Step 2: Adding Models to Test
Click "New Model" in the Testing tab and give it a name (e.g., "GPT-4", "Claude Sonnet", "Gemini Pro"). You can test as many models as you want and compare their results side-by-side in the Compare tab. Models are saved locally in your browser.
๐งช Step 3: Running a Test
1) Go to Pack Library and select a test pack
2) Switch to Testing tab, choose your model and tier
3) Click "Copy Test Prompt"
4) Paste into your AI model's chat interface
5) Copy the AI's complete response
6) Paste back into AIQ-X response box
7) Click "Analyze" to see results
๐ Step 4: Understanding Results
Scores measure response quality across multiple dimensions: depth, uncertainty calibration, structure, and domain-specific criteria. Higher scores indicate better capability in that area. Check the Compare tab for detailed analysis of strengths and weaknesses.
When the model completes its response simply select and copy the entire chat and paste into the app, only the actual response will be analyzed.
๐ Test Pack Library
Installed Packs
๐งช Testing Suite
Model Management
Select Test Pack
Select Test Tier
2. Choose your tier above
3. Copy the prompt
4. Paste into your AI
5. Copy full response back
6. Paste below and Analyze
Data Management
Latest Assessment Results
โ๏ธ Model Comparison
| Domain |
|---|
๐ฏ Model Recommendations
โ Help & Frequently Asked Questions
Q: What does AIQ-X measure?
AIQ-X evaluates cognitive patterns and capabilities like reasoning, creativity, self-awareness, and communication quality. Unlike traditional benchmarks that just check if answers are "correct," AIQ-X analyzes HOW models think and respond.
Q: Why did my model score low?
Low scores can indicate: (1) Response too brief, (2) Overconfident language (always/never), (3) Lack of reasoning structure, (4) Missing appropriate uncertainty. Scores measure response style, not absolute capability.
Q: Can I test ChatGPT, Claude, or Gemini?
Yes! Copy the test prompt, paste it into any AI chat interface, copy the AI's complete response, and paste it back into AIQ-X for analysis. Works with any text-based model.
Q: How do I import test packs?
Go to Pack Library โ Import Pack โ select the .json file. Or create custom packs using the Pack Builder.
Q: What's the difference between tiers?
Basic: 10-15 questions, ~5 min
Advanced: 20-30 questions, ~10 min
Expert: 40+ questions, ~20 min
Q: Is my data saved?
Yes, all data is saved in your browser's local storage. Nothing is sent to external servers. Use Export/Import to backup or transfer data between devices.