diff --git a/CHANGELOG.md b/CHANGELOG.md
index 4ef09ec..e46e15a 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,6 +1,6 @@
 # Changelog
 
-## 1.0.0 (prepared, not published)
+## 1.0.0 — 2026-09-10
 
 - Replace speculative parallel proposals with sequential adaptive Metropolis-within-Gibbs. Remove shared random-generator races and proposal-selection bias. Remove the thread-count argument from `AMWG::Init`; callers now pass only the starting state and log-posterior callback.
 - `Sample(n)` records exactly n completed sweeps; `Burn(n)` advances without retaining samples. Adaptation tracks sweeps independently of chain storage. Reinitialization resets the seeded generator and adaptation. Sampling no longer prints progress.
diff --git a/RELEASING.md b/RELEASING.md
index 5b4611e..b2bdded 100644
--- a/RELEASING.md
+++ b/RELEASING.md
@@ -1,6 +1,6 @@
-# Preparing the first versioned release
+# Releasing best-cpp
 
-Version 1.0.0 is prepared, not published. No prior release tags were present during this audit. This release changes historical sampling results, seeded sequences, threading behavior and error handling; read CHANGELOG.md before upgrading. Recompute previous inference results.
+Version 1.0.0 is the first versioned release. It changes historical sampling results, seeded sequences, threading behavior and error handling; read CHANGELOG.md before upgrading. Recompute previous inference results.
 
 1. Merge the sampler-correctness PR, then its packaging follow-up. Require Linux/macOS regression and sanitizer jobs, Windows/Linux/macOS CMake consumers, and the Bazel archive-consumer job to pass.
 2. Verify `CMakeLists.txt`, `MODULE.bazel`, `best/version.hpp`, the consumer module and changelog agree on 1.0.0. Run the commands in README.md from the reviewed commit. The archive test uses committed HEAD, so commit release files before running it.