MineralBids.com - Mineral Auctions & Marketplace
New tool - Experimental

Mineral Identifier AI

Upload a clear photo of a single specimen and let our experimental AI try to identify it. This tool is designed for clean specimen photos only. Photos taken through glass showcases, with labels, hands, clutter, or multiple specimens will often give incorrect results.
Works best with clear, well-formed specimens. Ambiguous or low-quality examples may give unreliable results.

Current model status: This tool is under active development and should be used as guidance, not as a final scientific identification. The model currently recognizes 134 classes, which is already near the practical limit for reliable visual identification. Some minerals - especially rare species or visually identical crystals - cannot be distinguished from photos alone. For this reason, the model focuses on common minerals and visually distinguishable varieties. Specimens that look identical, even to an experienced human, will also be indistinguishable to the AI.
Currently recognizes 134 classes

The current model supports the following minerals and varieties. Only a limited number of additional classes may be added, as visual identification has practical limits.

  • Caledonite
  • Acanthite
  • Adamite
  • Aegirine
  • Afghanite
  • Agardite
  • Amazonite
  • Analcime
  • Anatase
  • Anglesite
  • Annabergite
  • Apatite
  • Green Apophyllite
  • Aragonite
  • Arsenopyrite
  • Atacamite
  • Aurichalcite
  • Autunite - Metaautunite
  • Axinite
  • Azurite
  • Baryte
  • Benitoite
  • Beryl var. Aquamarine
  • Beryl var. Emerald
  • Beryl var. Goshenite
  • Beryl var. Heliodor
  • Beryl var. Morganite
  • Beryl var. Red Beryl
  • Boleite
  • Bournonite
  • Brazilianite
  • Brochantite
  • Brookite
  • Brucite
  • Calcite
  • Cassiterite
  • Cavansite
  • Cerussite
  • Chalcopyrite
  • Chrysoberyl
  • Chrysocolla
  • Cinnabar
  • Clinoclase
  • Conichalcite
  • Copper
  • Corundum var. Ruby
  • Corundum var. Sapphire
  • Creedite
  • Crocoite
  • Cubanite
  • Cumengeite
  • Cuprite
  • Cuprosklodowskite
  • Descloizite
  • Diamond
  • Diaspore
  • Diopside
  • Dioptase
  • Dolomite
  • Elbaite
  • Epidote
  • Erythrite
  • Ettringite
  • Euclase
  • Feldspar
  • Ferberite
  • Fluorite
  • Galena
  • Garnet
  • Goethite
  • Gold
  • Gypsum
  • Halite
  • Hematite
  • Huebnerite
  • Ilvaite
  • Kunzite
  • Kyanite
  • Legrandite
  • Linarite
  • Ludlamite
  • Malachite
  • Manganite
  • Marcasite
  • Mimetite
  • Muscovite
  • Neptunite
  • Non-mineral / Not a mineral
  • Olmiite
  • Orpiment
  • Peridot
  • Phosphophyllite
  • Plancheite
  • Prehnite
  • Proustite
  • Pyrite
  • Pyromorphite
  • Pyrrhotite
  • Quartz
  • Quartz var. Amethyst
  • Quartz var. Faden
  • Quartz var. Gwindel
  • Quartz var. Japan Law Twin
  • Quartz var. Mango
  • Quartz var. Prase
  • Quartz var. Rose Quartz
  • Quartz var. Smoky Quartz
  • Realgar
  • Rhodochrosite
  • Rhodonite
  • Rosasite
  • Rutile
  • Scheelite
  • Schorl
  • Serandite
  • Siderite
  • Silver
  • Smithsonite
  • Sphalerite - Gemmy
  • Sphalerite var. Marmatite
  • Spinel
  • Stibnite
  • Stilbite
  • Sulphur
  • Tanzanite
  • Tetrahedrite - Tennantite
  • Titanite
  • Topaz
  • Topaz - Blue
  • Torbernite Metatorbernite
  • Vanadinite
  • Vesuvianite
  • Vivianite
  • Wulfenite
Photo tips
  • One specimen only
  • Clean / plain background
  • Sharp, well-lit photo
  • No glass, reflections, labels, hands, or clutter
  • Well-formed, visually representative specimen (not broken, unclear, or ambiguous)
  • ⚠️ Low-quality or cluttered photos will likely produce wrong results.
Experimental AI
Ready for analysis
Upload an image to get a prediction. We do not permanently store uploaded images on the AI server. Results below 66% confidence are shown as Inconclusive.
⚠️ This tool is not designed for quick snapshots of cluttered or low-quality specimens

About this tool

We are building a powerful new tool for mineral identification, designed to continuously improve over time. However, this is not magic, and there are real limits to what image-based analysis can achieve.

This tool is currently experimental and does not guarantee correct results. It is a visual identification assistant, not a laboratory method.

Some minerals can appear nearly identical in photographs. Reliable identification in such cases may require physical or analytical testing, including hardness, streak, specific gravity, UV response, chemical analysis, Raman spectroscopy, X-ray diffraction (XRD), or scanning electron microscopy (SEM-EDS).