Same OPTIMADE query, six months apart: different results. Not a bug. MP, AFLOW, OQMD, and JARVIS-DFT update continuously without surfacing changes.

Without immutable snapshots, a published result is not reproducible.

alloybase.app/blog/posts/w...

Building an OPTIMADE client: prototype is an afternoon. Production took weeks.

For industry: one engineer-day costs more than a year of Lab access. For academics: data plumbing shows up in publication rate, not payroll.

Four databases, four hull distances for SrTiO3. None wrong.

The source: Hubbard U values, elemental reference states, and phase space coverage. Three compounding methodological differences.

alloybase.app/blog/posts/p...

You pull data from Materials Project, then AFLOW, and the formation energies don't line up.

Enterprise answer: Citrine. Mid-market: Alloybase. OPTIMADE search, versioned datasets, source attribution. In free public beta now.

alloybase.app/blog/posts/a...

New post on querying all 13 OPTIMADE providers in one request.

asyncio.wait_for, not asyncio.gather. page_limit=100, always. Structure-only vs. computed-property providers — they're not equivalent.

Full code: alloybase.app/blog/posts/query-13-optimade-databases-one-request/

MP, AFLOW, OQMD, JARVIS-DFT, COD, NOMAD, Alexandria, MC3D, ODBX. Entry counts, methodology, and which workflows each handles best.

Includes a section on property namespace differences that silently break cross-provider queries.

Published: alloybase.app/blog/posts/o...

Materials Project, AFLOW, OQMD, or JARVIS-DFT: which one do you query first?

The answer depends on what you're measuring. We compared all four on coverage, DFT settings, and update cadence.

Full comparison → alloybase.app/blog/posts/m...

#MaterialsInformatics #ComputationalMaterialsScience #DFT

Alloybase is now in public beta.

Query 13+ OPTIMADE-compliant materials databases in a single request. Unified schema, source attribution on every row.

alloybase.app

#MaterialsInformatics #OPTIMADE #MaterialsScience