Sources / epoch-ai

Epoch AI Benchmarking Hub

healthydefault: independent

epoch.ai/benchmarks · trust rank 90 · adapter epoch-ai

Observations

1,159

append-only score facts

Raw snapshots

immutable fetched payloads

Last success

2026-07-14 00:15Z

end of last succeeded/partial run

Freshness

1.2 h

hours since last success

License & attribution

CC-BY 4.0 — the cleanest license among our sources. Cited as: Epoch AI, "AI Benchmarking Hub", https://epoch.ai/benchmarks.

What we ingest

Per-benchmark CSVs from benchmark_data.zip: Epoch-run evals (FrontierMath, GPQA Diamond, MATH Level 5, SimpleQA Verified, SWE-bench Verified, OTIS Mock AIME) via their Inspect harness with stderr, plus the Epoch Capabilities Index. Epoch-run rows land as independent; third-party _external files carry lower default trust.

Ingestion runs

last 3 of the run history

Started	Duration	Status	Key stats
2026-07-14 00:15Z	0.1 s	succeeded	docs 1 · promoted 0 · blocked 0 · scoreAdded 0 · unchanged 1
2026-07-14 00:08Z	4.5 s	succeeded	docs 1 · staged 1,174 · promoted 1,174 · blocked 0 · scoreAdded 39 · unchanged 0
2026-07-13 20:03Z	1.5 s	partial	docs 1 · staged 1,174 · promoted 0 · blocked 1,174 · scoreAdded 0 · unchanged 0

partial = run finished but some rows are blocked on identity review — nothing is published until every referenced model/agent/benchmark is resolved.

Raw snapshots

latest 2

Fetched	URL	SHA-256	Last re-fetch
2026-07-14 00:08Z	https://epoch.ai/data/benchmark_data.zip	2e894b3a975d…	2026-07-14 00:15Z
2026-07-13 20:03Z	https://epoch.ai/data/benchmark_data.zip	e26b90031c81…	2026-07-13 20:03Z

Snapshots are immutable raw payloads — every published number traces back to one. Re-fetching an identical payload bumps the re-fetch time instead of storing a duplicate. Scores from this source appear on benchmark profiles and model pages with this source linked.