$669 lab-grown reference breaks eight sessions at $651
CaratRadar's one-carat average moved 2.77% in a single session after holding flat since 29 September. RAPI's natural one-carat index eased 0.11% on the same morning, to 3,906.01.
By The Diamonds Desk
What changed
The flat run ended. CaratRadar's one-carat lab-grown average reads $669, up from the $651 Carat Capital's tape carried in each of its eight dated files since 29 September.
What it means
- If you quote lab-grown by the stone: A single session moved the reference 2.77%. Anything quoted off last week's figure is now 2.77% light.
- If you sell natural against lab: The natural one-carat index eased the same morning. The gap between the two narrowed from both ends.
- If you read listing counts: Active listings fell about 13% to 873,170. A thinner book moves an asking-price average more easily.
- What this is not: A wholesale price. This is an average of asking prices on certified retail listings, not a settled trade.
Key figures
| Lab-grown 1ct averagefrom $651 | $669 |
|---|---|
| One sessionCarat Capital's division | +2.77% |
| Active listingsfrom 1,003,013 | 873,170 |
| RAPI 1ct, natural−0.11% on the day | 3,906.01 |
Source: CaratRadar lab-grown listings page, read 7 October 2026 · asking-price average, 1.00ct, across certified listings
One session, eighteen dollars
A 2.77% step. $669 against $651 is 2.77% on Carat Capital's division. The prior mark is looked up in the 6 October tape file rather than taken from the page itself.
The page says so. CaratRadar's own sentence gives the date, the listing count and the average together, and it carries a market price index of $1,481 alongside them.
| Tape file | Lab-grown 1ct | Change |
|---|---|---|
| 7 October | $669 | +$18 |
| 6 October | $651 | unch. |
| 5 October | $651 | unch. |
| 29 Sep – 4 Oct | $651 | unch. in each of six files |
| 27 September | $650 | — |
Notes on this table
The run was counted by opening each dated tape file and reading its labgrown_diamond_1ct mark, never by quoting a prior file's prose. $651 appears in eight dated files: 29 and 30 September and 1, 2, 3, 4, 5 and 6 October. There is no 28 September file, so the run is eight entries rather than nine days, and it is stated as entries. The 27 September file reads $650 and the 26th also $650, so $651 is itself a step up from that level. The $669 above is the first change in the mark since 29 September.
The book got thinner
Down 129,843 listings. Active certified listings read 873,170 against 1,003,013 on the prior session, a fall of about 12.9% in one day.
Thin books jump. An average of asking prices moves when the cheapest listings leave as readily as when sellers raise prices. The count and the average moved together.
Natural eased on the same morning
RAPI slipped 0.11%. Rapaport's one-carat index reads 3,906.01 at its own 7 October point, against the 3,910.26 Carat Capital printed for 6 October.
Different instruments. RAPI is a per-carat asking average for D-H, IF-VS2 stones on one marketplace. CaratRadar's figure is a whole-stone retail listing average across retailers.
The ratio narrowed. Dividing one by the other gives 5.84 times, down from 6.01 times on the prior session, on two differently built references rather than one.
What to watch
- 8 October 2026Whether $669 holds for a second session or reverts. A one-day step on a thinning book is the shape that most often retraces.
- Through October 2026The active-listing count. A second large fall would say inventory is leaving the market rather than prices firming.
- Early November 2026Rapaport's October monthly release, which prices the natural side of the same ratio on a different instrument.
The story so far
Go deeper
What would change this call
A return to $651 or below. The reference held one level for eight consecutive tape entries before this, so a single session's step is the weakest form of evidence this series produces.
How the two references are built
CaratRadar calculates a lab-grown price from active certified listings across major online retailers and refreshes it daily, publishing a one-carat average, a market price index and the full listed range. It is an asking-price average on retail inventory. RAPI is the average asking price of the ten per cent lowest-priced round D-H, IF-VS2, GIA-graded stones on Rapaport Trade, quoted per carat. Neither is a record of settled trades, and the two are not built to be divided into each other; the ratio above is printed with that limit stated.
Method
The lab-grown page was fetched to raw bytes at caratradar.com, HTTP 200, 116,840 bytes, saved to newsroom/sources/2026-10-07/caratradar.html. Exact-digit greps on that file: 669 fourteen hits, 873,170 three, $1,481 three. The page's own sentence, read from the raw bytes: As of October 7, 2026, the lab grown market includes 873,170 active listings, a market price index of $1,481, a 1ct average of $669, and listed prices from $50 to $772,237. A first grep for 1481 without its comma returned zero hits and was re-run with the comma, which returned three; the figure is on the page and the first search was the wrong search. The RAPI point was read from rapaport.com, HTTP 200, 304,515 bytes, saved alongside, with exact-digit greps of 3906.01 one hit, 956.24 two, 1313.85 two and 18207.54 three. PRIOR MARKS, looked up and not derived: $651 from newsroom/tape/2026-10-06.json and 3,910.26 from the same file. The eight-entry flat run was established by opening each dated tape file from 29 September to 6 October and reading its mark, not by quoting the prior tape's prose; there is no 28 September file, so the run is counted and stated in entries rather than days. ARCHIVE, by entity and period: searched website/content/articles.json for $669, zero hits, and for 669, one hit, which is a-forty-six-fifty-one-off-indian-gold of 2026-09-04 and concerns an Indian gold price, not this reference. Ours of 2026-10-06, rapi-point-three-oh-adds-three-point-two, carried this reference at $651 for the SESSION OF 6 OCTOBER; this item covers the SESSION OF 7 OCTOBER at $669. Same instrument, different session, so ADVANCED, not a repeat. Derived figures with the division shown: 669 less 651 is 18, divided by 651 is 2.77%; 1,003,013 less 873,170 is 129,843, divided by 1,003,013 is 12.9%; 3,906.01 divided by 669 is 5.84, and 3,910.26 divided by 651 is 6.01. The 1,003,013 listing count is the figure recorded in this paper's 7 October tape notes for the prior session and is labelled as a prior-session figure, not re-fetched. No pull quote is carried.