Carat$711, $650, $511: one lab-grown carat, three prices
CaratRadar's 1-carat lab-grown average read $711 on 11 August. TheDiamondPrice puts fair market at $650 inside a $590 to $760 range. This paper carried $511 from a third page on 7 August.
Two public reference pages for a one-carat lab-grown diamond were read at the same hour this morning and returned $711 and $650.
The most-bought stone in America does not have a price, it has a range, and the trade has been quoting the middle of that range as though it were a fact. Nothing on either page read today is wrong.
CaratRadar is measuring offers and says so; TheDiamondPrice is estimating transactions and says so. The failure is at the counter, where a single figure gets repeated without the qualifier that makes it meaningful, and a 9.4% error on a $650 stone is most of a retailer's margin on it.
The practical instruction is to stop asking what a lab-grown carat costs and start asking which of the two questions the number answers. Until one of these pages publishes an ask-to-sale spread, a jeweller pricing a case should treat every published lab-grown reference as an upper bound and find the lower one in its own till receipts.
| Figure | |
|---|---|
| Caratradar, Asking Average (11 Aug) | $711 |
| Thediamondprice, Fair Market (Aug) | $650 |
| Stonealgo (7 Aug, Not Re-Read Today) | $511 |
ITwo pages, one stone, sixty-one dollars apart
CaratRadar, republished today, puts the one-carat average at $711 and states on the page that its figures are asking-price averages by shape, carat, colour and clarity, drawn from active certified listings across major online retailers and refreshed daily.
TheDiamondPrice, carrying an August 2026 date, puts the fair market price of a one-carat round at $650 and says most comparable stones sit between $590 and $760, on a benchmark of E to F colour and VVS clarity. The gap is $61, or 9.4% on the lower figure, derived by this desk.
One is what sellers ask. The other is what stones fetch.
The two are not in conflict, and that is the more useful finding. CaratRadar's $711 sits inside TheDiamondPrice's own $590 to $760 band, above its $675 midpoint and above its $650 central mark.
An asking-price average built from live listings should sit above a transaction estimate, because a listing is an offer and a sale is an agreement, and the distance between them is the discount a buyer negotiates.
What neither page publishes is that distance as a number, so a jeweller reading one of them has no way to know which side of the trade it describes.
This paper carried a third reference on 9 August, StoneAlgo at $511 dated 7 August, and that page has not been re-read today; taken with today's two, the published span for the same stone runs from $511 to $711, about 39%.
IIAn offer is not a sale
The span matters because of what this stone now is. Lab-grown took 61% of United States engagement centre stones in 2025, a figure this desk published on 30 July, which makes the one-carat lab-grown round the single most-bought diamond in the American market. It is also the one with no settled price.
Natural goods have the Rapaport list, published weekly on Thursdays, and the RapNet Diamond Index behind it. Grown goods have a set of independent aggregators, each computing a different quantity, none of them clearing a market, and none of them stating which side of the bid they stand on.
IIIThe tape carries the number the article prints
This paper's own instruments are inside the problem, and they do not agree with each other either. The newsroom tape carries a lab-grown one-carat reference of $711, the CaratRadar figure read today.
The site's own tape chip carries $475, footnoted on the page as the midpoint of the published CVD wholesale band for a one-carat stone and dated 30 July, and it is locked to the lab-grown price list it links to so that the two can never drift apart.
Those are not two versions of one number. One is a retail asking average read this morning and the other is a wholesale band midpoint published twelve days ago, and the distance between them, about 50%, is roughly what the trade takes between the two ends of this stone.
Both are now labelled on the page with the quantity and the date, which is the only defensible way to carry either. The two natural-diamond indices this desk also tries to carry are not in the same position, and this article originally said they were.
The RapNet Diamond Index publishes its index levels as plain text on its own public page, above the members-only wall that covers the price list; the 1.00-carat series stands at 3,881.34 at the July 2026 point this paper's tape carries, a monthly reading the publisher revises after posting.
The Zimnisky rough index returns no numeric reading at all, because its current level sits inside a chart image and is sold rather than printed, and the status word this paper carries against it is paywalled.
01Method · the desk’s arithmetic+
Two pages read live on 11 August 2026; the third is the figure this paper carried in its wire on 9 August, dated 7 August, and not re-read today. The three describe different quantities.
02Corrections1 note
Correction, 25 August 2026: this article as published said that RAPI and the Zimnisky rough index had both returned no numeric reading and that both printed unreachable. That was right about Zimnisky and wrong about RAPI. Retested on 25 August with an ordinary unauthenticated request, rapaport.com/diamond-price-index/ returns HTTP 200 and publishes its index levels in the page's own inline data, 3,881.34 for the 1.00-carat series, with no login and no membership. The bad claim came from reading the members-only wall over the weekly price list as though it covered the whole page, when the index levels sit above that wall in plain text. What was right as published: the Zimnisky half, whose current level is embedded in a chart image with no extractable figure and whose correct status word is paywalled rather than unreachable, and the article's central finding about the one-carat lab-grown span, which nothing here touches. This paper cannot reconstruct what the Rapaport page returned on 11 August and does not claim the page changed; the claim is corrected on the evidence available today. Found by this paper's standards desk on 24 August 2026.
AED 10.44 between two Dubai gold pages, and holding
Two Dubai gold pages moved together overnight and stayed AED 10.44 apart.
The trade, filed before the New York open.
Prices, tenders and the one story that moved the industry overnight. Ninety seconds.
Subscribe free →