Cut Performance Registry · 21 July 2026
Which diamond cut performs best, measured
Nine cuts, one protocol. Every figure traced from the published geometry of the cut as it ships, ray-traced at each material's own refractive index, and scored against the same line: Tolkowsky's 1919 ideal. Born Clear, measured.
The short answer
Measured on light performance rather than on reputation, the highest-scoring round brilliant in this registry is BL-Miliano, at a Global score of 93.28 in diamond, 4.68 points clear of Tolkowsky’s 1919 ideal cut. BL-Rafaello follows at 92.57 and BL-Girandola at 91.92. The modern round brilliant as it is commonly cut today, listed here as Brilliant 73, scores 89.52, just under a point ahead of the 1919 specification itself at 88.61.
The answer changes with the material. In moissanite the order reshuffles and BL-M5 leads at 95.13, the widest margin over the standard anywhere in the registry. In cubic zirconia BL-Miliano and BL-Rafaello finish level at 95.06 and 95.05. A cut is solved against a refractive index, so no single cut is the highest performing cut in every stone.
And no single cut leads on every axis. The most fire in diamond is BL-Girandola’s, at 85.4. The densest flash pattern is BL-Miliano’s, at 98.9 scintillation. The steadiest stone as it tilts away from face-up is BL-Rafaello, at 96.9, and it is the only cut in the registry with no deficit on any of the eight axes. If you want one figure, read Global. If you want to know what a cut is for, read the term it wins on.
These are cut designs, measured as geometry. Every figure comes from one stated ray-tracing protocol run on exact facet geometry at a fixed seed, so it reproduces exactly rather than approximately. What it deliberately excludes is the workshop: a finished stone’s polish and symmetry deviation belong to the cutter and the rough, not to the cut. Read them as engine measurements of a design, to roughly ±2 points. They are not a laboratory cut grade and not a substitute for an IGI, GIA or AGS report.
Four ways in
Everything here runs on one ray census, so any cut can be set against any other. Which set you want depends on what you are choosing between.
The registry
Sorted by Global within each material. The standard is marked. A lead of under half a point is not claimed as a win; a deficit is shown as a deficit, however small.
| Cut | Facets | Global | vs T57 | Brilliance | Fire | Leak | Tilt | Scint. | Verdict |
|---|---|---|---|---|---|---|---|---|---|
| BL-Miliano | 201 | 93.28 | +4.68 | 94.8 | 83.0 | 2.25 | 92.5 | 98.9 | Ahead |
| BL-Rafaello | 145 | 92.57 | +3.97 | 93.9 | 80.2 | 4.94 | 96.9 | 96.0 | Ahead |
| BL-Girandola | 93 | 91.92 | +3.32 | 93.1 | 85.4 | 4.42 | 93.1 | 87.3 | Ahead |
| BL-Vitruvio | 185 | 91.65 | +3.04 | 92.6 | 78.7 | 2.17 | 92.5 | 98.4 | Ahead |
| T105 | 105 | 90.55 | +1.95 | 95.2 | 76.6 | 1.38 | 87.3 | 90.3 | Ahead |
| T73 | 73 | 89.85 | +1.24 | 92.9 | 78.3 | 1.91 | 92.2 | 80.3 | Ahead |
| Brilliant 73 | 73 | 89.52 | +0.91 | 94.0 | 78.0 | 3.53 | 89.3 | 80.3 | Ahead |
| T57 standard | 57 | 88.61 | 0.00 | 92.3 | 78.8 | 1.88 | 90.4 | 71.8 | Benchmark |
| BL-Anteros | 73 | 88.39 | −0.22 | 94.6 | 74.2 | 2.30 | 84.9 | 80.3 | Behind |
| BL-M5 | 85 | 87.79 | −0.81 | 96.7 | 66.9 | 2.57 | 83.2 | 84.9 | Behind |
| Cut | Facets | Global | vs T57 | Brilliance | Fire | Leak | Tilt | Scint. | Verdict |
|---|---|---|---|---|---|---|---|---|---|
| BL-M5 | 85 | 95.13 | +5.31 | 98.1 | 99.6 | 1.05 | 84.3 | 84.9 | Ahead |
| BL-Vitruvio | 185 | 92.86 | +3.04 | 86.1 | 97.2 | 7.46 | 94.0 | 98.4 | Ahead |
| BL-Miliano | 201 | 92.31 | +2.49 | 86.6 | 92.5 | 7.58 | 95.1 | 98.9 | Ahead |
| T105 | 105 | 91.67 | +1.85 | 86.1 | 96.7 | 5.64 | 91.1 | 90.3 | Ahead |
| T73 | 73 | 91.04 | +1.22 | 86.0 | 97.2 | 7.40 | 94.1 | 80.3 | Ahead |
| BL-Anteros | 73 | 90.93 | +1.11 | 91.9 | 91.1 | 2.76 | 85.7 | 80.3 | Ahead |
| BL-Rafaello | 145 | 90.71 | +0.89 | 84.2 | 90.3 | 12.46 | 97.8 | 96.0 | Ahead |
| BL-Girandola | 93 | 90.49 | +0.67 | 84.0 | 93.0 | 7.90 | 96.2 | 87.3 | Ahead |
| Brilliant 73 | 73 | 89.93 | +0.11 | 86.5 | 95.8 | 7.44 | 88.0 | 80.3 | Level |
| T57 standard | 57 | 89.82 | 0.00 | 85.3 | 97.5 | 7.76 | 93.2 | 71.8 | Benchmark |
| Cut | Facets | Global | vs T57 | Brilliance | Fire | Leak | Tilt | Scint. | Verdict |
|---|---|---|---|---|---|---|---|---|---|
| BL-Miliano | 201 | 95.06 | +3.28 | 97.1 | 89.9 | 1.17 | 89.2 | 98.9 | Ahead |
| BL-Rafaello | 145 | 95.05 | +3.28 | 99.7 | 87.1 | 0.13 | 88.2 | 96.0 | Ahead |
| BL-Vitruvio | 185 | 94.59 | +2.81 | 98.4 | 87.0 | 0.58 | 87.1 | 98.4 | Ahead |
| BL-Girandola | 93 | 93.88 | +2.10 | 97.8 | 91.6 | 0.51 | 85.2 | 87.3 | Ahead |
| T105 | 105 | 93.33 | +1.56 | 98.4 | 85.2 | 0.90 | 86.9 | 90.3 | Ahead |
| T73 | 73 | 92.77 | +1.00 | 98.4 | 86.8 | 0.60 | 87.4 | 80.3 | Ahead |
| Brilliant 73 | 73 | 91.92 | +0.15 | 97.1 | 84.6 | 1.19 | 88.1 | 80.3 | Level |
| T57 standard | 57 | 91.77 | 0.00 | 98.0 | 86.8 | 0.51 | 87.1 | 71.8 | Benchmark |
| BL-Anteros | 73 | 88.98 | −2.79 | 92.3 | 81.7 | 6.55 | 86.9 | 80.3 | Behind |
| BL-M5 | 85 | 87.12 | −4.65 | 91.1 | 73.3 | 8.68 | 86.7 | 84.9 | Behind |
The Index draws each stone as a real-time 3D render, and the render simplifies the facet pattern so it can be traced live in a browser. No figure on this page comes from it. Every number here is computed from the cut’s full facet geometry on a fixed seed, so it reproduces exactly rather than approximately.
What the tables say
Six of the nine cuts we measure against the standard beat it in all three materials, and that is the first thing to understand about this table. T57 is a 57-facet design from 1919. It carries the fewest facets of anything published here, and scintillation on this scale is a direct function of facet count, so almost any modern cut clears it. The question worth asking is never whether a cut beats the standard. It is how much of the margin is optics and how much is just more facets.
The T-series answers that question directly. T57, T73 and BL-Vitruvio are one design at three facet counts: the same 1919 angles, cut finer. Across all three, brilliance moves 0.6, fire 0.5 and leak 0.3, while scintillation climbs 71.8 → 80.3 → 98.4. Of BL-Vitruvio’s 3.04-point lead over the standard, 2.65 (87% of it) is facet count alone. Read every other row against that.
BL-Miliano is the strongest all-rounder, ahead in diamond (93.28), cubic zirconia (95.06) and moissanite (92.31), and the only cut ahead on brilliance and fire at the same time in diamond. BL-Rafaello gives up nothing on any of the eight axes: six clear wins, symmetry level at the 100 ceiling, and spread level within a tenth of a point. It also holds the steadiest stone under tilt (96.9 in diamond, 97.8 in moissanite) and the cleanest light budget in zirconia, at 0.13% leak. BL-Girandola carries the most fire of the house diamond cuts, at 85.4.
Measured against BL-Vitruvio rather than against a 57-facet stone from 1919, the three house cuts stand at +1.63 (Miliano), +0.92 (Rafaello) and +0.27 (Girandola) on Global, and +1.61, +1.17 and +1.39 on the optical terms alone. BL-M5 is the fourth house cut and belongs in the moissanite column, where it stands +2.27 above Vitruvio and +3.62 on the optical terms, the largest optical margin in the registry in any material. Those are the margins that were designed rather than counted. Each cut earns its own on a different axis, and in every one of them the angles and the facet pattern are doing opposite jobs: the angles buy one property and spend another, and the tiling pays the difference back. Miliano’s shallow crown gives away fire and its 201 facets return it. Rafaello’s angles buy fire and cost stability; its 145 facets buy the stability back. Girandola’s angles buy the fire and lose the light; its 93 facets recover it. Each cut page sets out which half did what.
And none of them beat the ideal by leaving his pavilion. Swept on his own 57-facet architecture, Tolkowsky’s 40.75° sits 0.05 of a point off the optimum, and the optimum tracks the refractive index exactly as Diamond Design says it should: 41.25° in cubic zirconia, 40.25° in diamond, 38.25° in moissanite, a 3° range. Hold the index and change the facet layout instead and it moves about a degree, a third as far. Every house cut here sits within three quarters of a degree of his pavilion. Where they depart is the crown, by up to three degrees, and the crown is the parameter he derived from appearance rather than from the index. On this engine Global holds its value across 0.75° of pavilion and 2.5° of crown. The cuts leave the pinned parameter alone and move the loose one.
BL-M5 is the moissanite specialist, at 95.13, with 99.6 fire and 1.05% leak, and the only cut that finishes behind the 1919 ideal in both diamond and zirconia. It is cut for one material and shows it, in both directions.
The standard
Every number on this page is measured against a specification Marcel Tolkowsky published in 1919: a 53% table, a 34.5° crown and a 40.75° pavilion, worked out with trigonometry rather than tradition. It is worth knowing what he solved for, what he left alone, and how the four cuts in the T-series each change one thing about it.
The cuts
Common questions
On measured light performance, the highest-scoring round brilliant in this registry is BL-Miliano, at a Global score of 93.28 in diamond, 4.68 points above Tolkowsky’s 1919 ideal. BL-Rafaello follows at 92.57 and BL-Girandola at 91.92. The modern round brilliant as it is commonly cut today scores 89.52, just under a point ahead of the 1919 specification itself. Best is also material-dependent: in moissanite the leader is BL-M5 at 95.13, and in cubic zirconia BL-Miliano and BL-Rafaello finish level at 95.06 and 95.05. These are Brilliani Labs’ own engine measurements, not a laboratory grade.
The ideal cut is the round brilliant specification Marcel Tolkowsky published in Diamond Design in 1919: a 53% table, a 34.5° crown angle, a 40.75° pavilion angle and 59.3% total depth on 57 facets. It is still the line the trade ships against, and it is the benchmark every cut in this registry is measured against. Grading laboratories now call the top step Excellent rather than Ideal, but they are describing the same target. The 1919 ideal, in full.
Sparkle splits into three measurable parts, and a different cut leads each one. For scintillation, the flash pattern as the stone moves, BL-Miliano leads at 98.9 in diamond, because scintillation rises with active facet count. For fire, the coloured flashes, BL-Girandola leads at 85.4. For sparkle held onto when the stone tilts away from face-up, BL-Rafaello leads at 96.9. No single cut leads on all three.
Not on the Global score. Seven of the nine cuts measured against it finish ahead in diamond, and six of the nine finish ahead in all three materials. The reason is mostly facet count rather than better angles: T57 carries the fewest facets of anything published here, and scintillation is a direct function of active facet count. His pavilion angle survives the test almost intact. Every house cut here sits within three quarters of a degree of his 40.75° and departs on the crown instead.
BL-Girandola, at a fire figure of 85.4 in diamond, the highest measured in the registry. In moissanite the highest fire figure is BL-M5 at 99.6, because moissanite disperses light more strongly and rewards a cut solved for that index.
Only on one term. T57, T73 and BL-Vitruvio are the same 1919 angles cut at 57, 73 and 185 facets. Across that range brilliance moves 0.6 points, fire 0.5 and leak 0.3, while scintillation climbs from 71.8 to 80.3 to 98.4. So 87% of BL-Vitruvio’s 3.04-point lead over the standard is facet count alone. More facets buy a finer flash pattern, not more returned light.
BL-M5, at 95.13 in moissanite, 5.31 points above the 1919 ideal and the widest margin in the registry in any material. It is solved for one refractive index and shows it in both directions: it is the only cut here that finishes behind the standard in both diamond and cubic zirconia.
Monte Carlo ray tracing on the exact facet geometry of each cut: 3,200 rays face-up plus 2,200 at 20° of tilt, cosine-weighted from a 7° near-vertical cone, at a fixed seed so the figures reproduce exactly rather than approximately. The Global score is 0.34 useful light + 0.20 fire + 0.16 tilt + 0.10 scintillation + 0.10 symmetry + 0.10 (100 minus leak). What is measured is the cut design rather than a finished stone, because polish and symmetry deviation belong to the cutter rather than to the geometry. The engine’s check is Tolkowsky’s own 40.75°: swept on his 57-facet layout, this engine reproduces his figure to within 0.05 of a point. Read the results as engine measurements of roughly ±2 points, not laboratory-certified grades.
None of them beats a round brilliant for light return, and on one protocol the gap is measurable. Of the twenty-one jewellery shapes measured in the jewellery cuts section, the plain square step cut gives up least at 84.11 Global in diamond, then the cushion at 83.68 and the trillion at 80.86, against the standard’s 88.61. The oval reads 74.96, the princess 76.20 and the emerald 70.48. A shape is chosen for how it looks, so the useful question is not which is best but what each one costs you.
The documented ancestry of the round brilliant runs: the single cut from the 1300s at 18 facets, the Mazarin double cut of the mid-1600s at 34, the Peruzzi triple cut of the early 1700s at 58, the old mine cut that dominated until the late 1800s, and the old european cut of roughly 1890 to 1930, which lands 0.78 points from the 1919 ideal. All five are rebuilt from Tolkowsky’s Diamond Design and measured on this protocol in the historic cuts section. The finding is that three centuries of adding facets barely moved light return; mechanical bruting, which made stones truly round, moved all of it.
A shape is the outline of the stone: round, oval, pear, cushion, emerald. A cut design is the facet plan inside that outline, and a cut grade is how well that plan was executed. Every cut in this registry is a round brilliant, so they differ from each other in facet plan rather than in silhouette. The outlines are measured separately in the jewellery cuts section, on the same protocol, so the two sets can be read against each other.
Method
Monte-Carlo on exact facet geometry: 3,200 rays face-up plus 2,200 at 20° tilt, cosine-weighted from a 7° near-vertical cone. Brilliance is the light that exits through the crown within 64° of vertical.
0.34 useful + 0.20 fire + 0.16 tilt + 0.10 scintillation + 0.10 symmetry + 0.10 (100 − leak). Every cut scored against T57, Tolkowsky's 1919 ideal as he specified it, in the same material. The T-series shares that stone's angles and varies only the facet count.
A cut is a geometric specification, so that is what is measured here. Polish deviation, symmetry error and inclusions belong to the cutter and the rough, not to the design; folding them in would measure the workshop instead of the cut. Diamond Design works from the same premise, and Tolkowsky measured no stones. Symmetry therefore reads 100 for every entry, a constant term that cancels out of every comparison.
Not a cut stone. Tolkowsky derived 40.75° analytically by hand in 1919; sweeping pavilion angle on his own 57-facet layout puts this engine’s diamond optimum 0.05 of a point away from it, and the optimum tracks refractive index across three materials exactly as Diamond Design says it should. Two unrelated routes to one figure is the validation that means something for geometry. Figures are reproducible rather than approximate: the census runs on a fixed seed. Treat them as engine measurements to roughly ±2 points, not as a laboratory cut grade.
A facet count here is the planes that actually form a face. The denser sets approximate the girdle with a ring of planes, and any that end up buried inside the hull are left out, because a buried plane throws no flash. Tilt is a retention figure, the share of face-up return still delivered at 20°, capped at 100.