diff --git a/README.md b/README.md
index f10a5ec..7107a69 100644
--- a/README.md
+++ b/README.md
@@ -159,7 +159,7 @@ item can be highlighted in the original advisory.
 | 3 · Neural models | Frozen leakage-safe splits, annotation batches (01–03, 06: model-drafted), DistilBERT and DeBERTa-v3 with span / role / group heads, experiment runner, CRF ablations and learning curve | 🟨 trained on model drafts (run 3) · DeBERTa ≈ CRF on held-out agreement, no clear winner |
 | 4 · Reliability | Calibration and review policy, formatting robustness (rules: 100%), challenge-set and error-catalogue tooling | 🟨 tooling ready · thresholds await labels |
 | 5 · Application | FastAPI API, review UI, SQLite (WAL) with append-only history, exports, Docker Compose stack, release manifests, tested rollback, CI image smoke test | ✅ built |
-| 6 · Production | Benchmark harness, ONNX/INT8 export (FP32 parity exact), local proxy benchmarks and load tests, monitoring; CRF v0.1.0 measured on the 2 vCPU / 4 GB target host | 🟨 CRF meets every serving target on the target host · transformers not yet measured there |
+| 6 · Production | Benchmark harness, ONNX/INT8 export (FP32 parity exact), local proxy benchmarks and load tests, monitoring; CRF v0.1.0 measured and deployed (`app verify` passed) on the 2 vCPU / 4 GB target host | 🟨 CRF meets every serving target on the target host · transformers not yet measured there |
 
 ### Preliminary pilot results (dev split only)
 
diff --git a/docs/deployment.md b/docs/deployment.md
index ee06b80..13acda7 100644
--- a/docs/deployment.md
+++ b/docs/deployment.md
@@ -80,6 +80,20 @@ save (which must not duplicate) and export, and reads the saved review back.
 For transformer releases, build the image with `--build-arg EXTRAS="serve neural"`. uv
 installs CPU-only PyTorch from the lockfile.
 
+### Hosted deployment record
+
+**2026-11-09, AWS EC2:** release v0.1.0 (CRF `crf-batches-01-02`, image
+`sha256:0b9e2f30…adac`) was deployed with the steps above on a 2 vCPU / 4 GiB x86-64 instance
+running Ubuntu 24.04, behind Caddy with automatic HTTPS (an `sslip.io` name). `triggermap app
+verify` passed every check:
+- `/readyz` matched `release.json`;
+- extraction of the bundled sample returned 4 conditions;
+- a review was saved, and its retry did not duplicate it;
+- the review was retrievable and exported.
+
+The same host produced the [target-host benchmarks](benchmarks.md). The instance was a
+temporary test deployment and was shut down afterwards; there is no permanent public instance.
+
 ## Scheduled refresh
 
 Advisories change. Refresh the snapshot daily from the host's crontab: