AI WRITING COMPARISON WORKSHEET
AI Undetectable | 2026-09-18
Guide: https://aiundetectable.com/posts/chatgpt-vs-claude

This is a blank worksheet, not measured results or a model ranking.
Keep submitted text and full outputs privately with your own records.

TEST SETUP
Date (UTC):
Reader and intended use:
Exact model names and versions:
Generation settings, if available:
Source facts that both drafts must preserve:
Exact shared prompt:
Target length:
Selection rule: retain the first complete response from each model.

SAMPLE RECORD (duplicate for each output)
Sample ID:
Assistant / exact model:
Genre (article, essay, customer email, or known human control):
Source brief filename:
Full output filename:
Actual word count:
Missing required facts:
Invented claims:
Reader-fit observations:
Paragraph structure observations:
Factual repairs required:
Substantial edits required:
Editing time in minutes:

OPTIONAL DETECTOR RECORD (separate from writing quality)
Detector name and version, if shown:
Check date (UTC):
Exact submitted output filename:
Score and its label exactly as displayed:
Human control provenance and permission:
Disagreement with other detectors:

COMPARISON SUMMARY
Number of outputs tested per model and genre:
Missing-fact count per model:
Invented-claim count per model:
Editing effort and reader-fit differences:
Detector results, including inconvenient results and human controls:
Limitations of this small sample:
Which model fits this specific workload, and why: