Provenance Record

Well-structured News piece (658 words, within the 400-1200 range). Clear sectioning (Overview / What We Know / What We Don't Know / Analysis) with appropriate technical depth for a release-coverage article. Every paragraph anchors its specifics to an inline-linked source.

All 5 source snapshots read from disk and verified (all status 200, no archive fallback). source-0 (valkey.io blog): confirms May 19 2026 release, 'over eighty contributors', database-level ACLs scoping user permissions to specific databases, Lua extracted to a module reducing security surface area, automatic background TLS cert reloading + INFO expiration dates, JSON logging via the `log-format json` directive, new cumulative main/I-O thread usage metrics, I/O threading redesign 'up to 17%', string GET 'up to 30%', XRANGE/XREVRANGE 'up to 30% faster', hardware clock default GET/SET 'up to 3%', 'up to 20%' memory reduction for strings under 128 bytes, 'up to 10%' sorted-set memory reduction, and new commands HGETDEL/MSETEX/CLUSTERSCAN — all verbatim-supported. source-1 (Linux Foundation): confirms the first Madelyn Olson quote verbatim and her exact title 'Valkey project maintainer and Principal Engineer for AWS in-memory databases', the 'per-key memory usage by up to 10% for common workloads' with no tuning framing, the 'database-level ACL system, providing fine-grained multi-tenant isolation within a single instance' phrasing, and the ecosystem facts (Valkey Search 1.2, Valkey Admin GA visual cluster management tool, Valkey GLIDE 2.4 with C#/PHP). source-2 (PRNewswire): verbatim mirror of the LF release; independently confirms the Olson quote and the 10% memory figure. source-3 (diginomica): confirms the title, '19 May' ship date, the 10% per-key memory reduction, the 'Linux Foundation-governed open source Redis fork'/vendor-neutral framing, the second Olson quote verbatim ('Valkey has been using AI agents to automate tedious, time-intensive tasks now, while the wider industry is working on large-scale agentic models'), and the durability/ACID 'road to 10.0' roadmap that the article paraphrases (outside quote marks) as 'full ACID compliance with fast failover is targeted for version 10.0'. source-4 (github.com/valkey-io/valkey/releases): confirms '9.1.0 Latest ... Upgrade urgency LOW: This is the first stable release of Valkey 9.1.' Every direct quote is verbatim and correctly attributed to the correct speaker.

No hallucinations detected. Both Olson quotes are exact and correctly speaker-attributed with the full title. All performance/memory percentages, the release date, contributor count, ACL/JSON/threading claims, ecosystem version numbers, and the GitHub upgrade-urgency LOW status trace to cited snapshots. The unsupported 'roughly eight months after 9.0' interval was correctly removed (absent from the body). The Valkey blog's '2.1 million requests/sec at 512-byte payloads' headline benchmark was correctly DROPPED — it appeared only as a Phoronix figure elsewhere (Phoronix 403'd/bot-blocked, single-sourced) and the article deliberately closes by noting the release targets operational concerns 'rather than at headline benchmark figures'. The internal cross-reference to /article/2026-03/02-valkey-turns-two-... resolves to an existing published article.

High-quality, accurate, neutral release-coverage submission with all five sources personally read and verified, both quotes verbatim, all specifics sourced, the speculative benchmark and unsupported interval correctly dropped, and the internal cross-reference resolving. The only issue is the diginomica.com allowlist gap, which is honestly summarizable in a corrections note. Verdict: APPROVE_WITH_CORRECTIONS.