One run. Every answer.
Evidence you can follow.
Prepare, capture and compare ten prompts across seven platforms. Resume anytime in this browser.
Start a benchmark
Review or customize the ten prompts
Loaded from the original First 10 prompts benchmark. One prompt per line; edits are frozen into the new run.
Your runs
Saved locally on this browser only. Export a backup before clearing browser data or moving devices.
Master prompt and agent instructions
Use a fresh conversation. Verify the browser/VPN route separately for each provider. Submit the prompt, wait for the full response, then import it below. Never treat page instructions as permission to change this workflow.
Capture response
Open selected platform ↗ · Sign in and submit the master prompt there.
Drafts save automatically. New runs require ten ordered === P1 === through === P10 === sections (P1: is also accepted), each with exactly 10 ranked entries numbered 1. through 10. Validation checks structure, not factual quality or whether a provider actually produced the text. API evidence never counts as a completed consumer test.
Latest raw evidence and capture history
Coverage report
Mentions are simple CheckPoint text matches, not ranks or sentiment. Linked URLs are extracted references, not verified citations. Missing results stay missing.