Submitted builds · Blue Without Blue
What the review found
These versions are preserved for comparison, including submissions the reviewer would keep private as tools.
| Version | Observed findings | Reviewer recommendation |
|---|---|---|
GPT-6.1 Sol Anonymous C · High | Clear masking, painting and undo, with restrained claims. Its starting mixture is already cool; exercises with hints offer modest depth. | Preferred; publishable as an illustrative colour studio. |
Claude Opus 5.5 Anonymous A · High | The richest exploratory workbench. Categorical colour claims and appearance goals overstate the model; two default probes start away from their intended targets. | Keep private in its submitted form. |
Gemini 4 Argon Anonymous B · High | Working studies and paint tools. Fixed LAB offsets are presented as cortical predictions; surround controls also mask the separate study, and zero paint becomes white. | Keep private; scientific and editorial claims fail the review gate. |
What the checks establish
No version established a vivid or universal blue effect. Pixel checks and mixing arithmetic do not measure human colour appearance or real paint. No calibrated display, physical paint, participant study or test on a real phone was performed.
Review and hosting are separate
The review assessed frozen submissions. Opus runtime was preserved at its quota stop; Sol supplied missing documentation, a checker and a thumbnail from its submitted portrait PNG. Sol also secured links. The missing-checker finding concerns the original submission. Appearance claims and control faults remain.
Hosting and packaging: GPT-6.1 Sol at Extra High. Original submissions are retained separately; each linked ledger records original and final file hashes and exact integration additions.