Provenance Record

Well-structured News piece (877 words, within the 400-1200 News range). Clear sectioning: Overview, What We Know, The Benchmarks, Business Context, and a What We Don't Know section that explicitly hedges the vendor-run benchmarks. Technical depth (NVMe colocation, native CDC via ClickPipes, pg_clickhouse query pushdown, pgbench/TPC-H methodology) is accurate and appropriately scoped.

Read all 5 gzipped snapshots from disk (source-0 through source-4); sha256 of each decompressed file matches manifest.json exactly. ALL 5 sources are ClickHouse first-party blog posts (vendor); all returned HTTP 200 with no archive fallback. source-0 (beta announcement): verified verbatim — 'remains free until usage metering begins on June 15, 2026', 'up to 10x faster transactional performance', 'During Beta, all plans include a 50% discount', '$300 in free credits', '1 vCPU / 8 GB RAM / 59 GB NVMe ... starting at approximately $32/month', 'Native CDC via ClickPipes and the pg_clickhouse extension are included at no additional cost', and the GA hedge 'As we move toward General Availability, pricing and packaging may evolve'. source-1 (private preview): verified verbatim — definition quote, NVMe 'physically colocated with compute ... EBS-backed volumes', pg_clickhouse 'comprehensive query pushdown ... FILTERs, JOINs, SEMI-JOINs', 'up to 100X faster analytics', '14 of 22 TPC-H queries ... over 60x', date 'Jan 22, 2026'. source-2 (Open House recap): verified verbatim — 'from private preview to public beta', native CDC pipeline quote, 'over 5x more transactions per second than AWS RDS and 2.4x more than the next closest alternative', TPC-H '54.7 seconds on eight nodes, down from 117.6 seconds on a single node'. source-3 (PostgresBench): methodology verified (pgbench, 256 clients, 16 threads, 'Each run lasts 10 minutes', ~100 GB scale factor); all five per-vendor TPS figures confirmed in the data table at the Large config — ClickHouse 28668, AWS RDS 8133, AWS Aurora 12628, Neon 8563, Crunchy Bridge 14790 — each correctly attributed; 'reproducible benchmark' and 'full benchmark repository is open source' confirmed. source-4 (ARR announcement): Aaron Katz CEO quote verified verbatim, '$250 million in annual run-rate revenue', 'more than 1,000 net new customers since January, bringing its total to 4,000', and 'May 27, 2026' / 'SAN FRANCISCO' Open House dateline confirmed. Every direct quote, every numeric specific, and every date traces to its cited snapshot.

No hallucinations detected. Every claim maps to a cited source. The article correctly attributes each performance figure to the specific source that published it (the TPS table to PostgresBench, the 5x/2.4x summary to the Open House recap, the analytics 100X/60x figures to the private-preview post). The Databricks Lakebase internal cross-reference (/article/2026-04/12-databricks-lakebase-...) resolves to an existing published article and is not duplicated.

High-quality, accurately sourced News submission. Despite single-vendor sourcing, the article maintains neutrality through consistent attribution and an explicit limitations section, and every quote, number, and date verifies verbatim against the committed snapshots. Integrity checks pass (signature and hash valid). APPROVE.