Inferrex

/ claims & stats

Every number, sourced and dated.

Marketing decks round up. We don't. Every figure on this page resolves live from the platform's own stats manifest — current as of 4 September 2026. Where a number is a placeholder, projected, or blocked on a pipeline step, it says so.

The corpus — discovered, specced, curated.

A funnel: from the known vendor universe, down to providers with a full machine-read specification, down to the curated connector library. Each row is a tighter, deeper cut of the one above it.

234,373
Vendors discovered

As at 4 September 2026

1,941
Providers fully specced

As at 4 September 2026

8314
Curated connectors

As at 18 June 2026

25,359
API specs ingested

As at 4 September 2026

38
Schema formats normalised

As at 18 June 2026

17
Semantic types

As at 18 June 2026

Classified, not just collected.

Every spec is parsed into the canonical model — entities, fields, and the integration pairs they make possible.

1.5 million
Entities classified

As at 4 September 2026

7,194,438
Fields read and catalogued

As at 4 September 2026

21.1 million
Integration pairs

As at 4 September 2026

Cloud coverage.

Every major cloud — AWS, Azure, GCP, and Oracle Cloud — parsed into the canonical model and ready for cross-cloud migration pairs. The per-cloud spec counts are in the chart above; the totals below.

23,623
Cloud specs classified

As at 4 September 2026

The proprietary AI stack — honestly.

Layers are built first, then promoted only after passing the evaluation harness — so 'built' and 'promoted' are reported separately, live from the manifest below.

14
AI layers built

As at 18 June 2026

11
Adapters promoted

As at 18 June 2026

13
Layers in training

As at 18 June 2026

864k
Training samples

As at 18 June 2026

464
Golden mappings

As at 18 June 2026

44
Golden-mapped providers

As at 18 June 2026