Context
XGrade compared with a typical AI pre-grade and a slab house.
| XGrade | Typical AI pre-grade | PSA / BGS / CGC / SGC / TAG | |
|---|---|---|---|
| What you send | Front (+ back) photos | Usually one photo | The physical card |
| What you get | Visible-condition report + published math | A predicted company number, often with a % claim | An official grade in a holder |
| Who picks the number | A deterministic formula you can read | The model, or an unpublished blend | A human grader (company process) |
| Unseen corner | Scores 7. Not assumed gem. | Often still a high number | They see the corner |
| No back photo | Max 9, printed | Often still a 10 forecast | They see both sides |
| Protocol | Fully published | Usually a marketing page | Company standards, not your formula |
| Possession | Never | Never | Yes, until return |
| Affiliation | None | None, despite the prediction labels | They are the company |
| Time / cost | Minutes; model usage billed by xAI | Seconds; subscription or credits | Weeks to months; service + ship + insurance |
Why we will not quote 92.8%
A photo-only system cannot know the physical card. Publishing a single accuracy percent implies the opposite. XGrade publishes caps and evidence instead. If the photos are incomplete, the number is capped in writing.
When you still need a slab house
When you need an official grade, a holder, population reports, or marketplace convention. Use XGrade first so you do not pay to learn the photos already showed an 8.