Local workspace · No model calls
Same-task comparison lab
Compare outputs you actually collected against a shared synthetic task. Record provenance and judge criteria manually.
1. Use the same task
Run the exact prompt and input yourself in the tools you choose. Paste each output below, label its origin and record the model, run date and settings when known. No requests are sent to any model here. Comparing two user-selected outputs is not a controlled benchmark.
delivery-notice · version 1 · invented demonstration input
Prompt
Write a customer update of at most 60 words using only the supplied facts. Preserve uncertainty and do not invent tracking or refund terms.
Input
Order DEMO-17 was dispatched on Monday. Delivery is estimated for Thursday, not guaranteed. Tracking is unavailable. No refund policy is provided.
2. Output A
A demonstration is text you wrote to illustrate a response. A user-recorded output is one you actually obtained elsewhere; Moyan has not verified it.
0 / 8,000 characters
0 / 120 characters
0 / 10 characters
0 / 120 characters
Your manual judgments
Editing output text clears its judgments. No scores or winner are calculated.
2. Output B
A demonstration is text you wrote to illustrate a response. A user-recorded output is one you actually obtained elsewhere; Moyan has not verified it.
0 / 8,000 characters
0 / 120 characters
0 / 10 characters
0 / 120 characters
Your manual judgments
Editing output text clears its judgments. No scores or winner are calculated.
3. Keep your record
Export includes the versioned fixture, your outputs, unknowns and manual judgments. There is no import or automatic saving.