diff --git a/src/openhuman/inference/provider/factory_test.rs b/src/openhuman/inference/provider/factory_test.rs
index 3706b00e1..b4d9abe37 100644
--- a/src/openhuman/inference/provider/factory_test.rs
+++ b/src/openhuman/inference/provider/factory_test.rs
@@ -760,8 +760,9 @@ fn make_openhuman_backend_forwards_unknown_hint_verbatim() {
     // Unrecognised hint:* strings (e.g. hint:reaction for lightweight models)
     // must be forwarded to the backend unchanged. The backend is authoritative
     // over which hint values it accepts; the factory only translates the
-    // canonical hints (reasoning/chat/agentic/coding/summarization) — the last
-    // of which gained a dedicated tier, so it is no longer "unknown".
+    // canonical hints (reasoning/chat/agentic/coding/summarization).
+    // `hint:summarization` became canonical when `summarization-v1` shipped
+    // (PR #2690), so it is no longer a passthrough case.
     for hint in ["hint:reaction", "hint:garbage"] {
         let mut config = Config::default();
         config.default_model = Some(hint.to_string());
diff --git a/src/openhuman/memory/chat.rs b/src/openhuman/memory/chat.rs
index 900a3cfa0..4afa381e2 100644
--- a/src/openhuman/memory/chat.rs
+++ b/src/openhuman/memory/chat.rs
@@ -215,9 +215,9 @@ mod tests {
     fn build_chat_runtime_defaults_to_openhuman_resolved_model() {
         let cfg = Config::default();
         let (_provider, model) = build_chat_runtime(&cfg).unwrap();
-        // build_chat_runtime resolves the "summarization" role, which gained a
-        // dedicated tier (summarization-v1) — the role no longer falls back to
-        // the generic reasoning-v1 default.
+        // build_chat_runtime resolves the "summarization" workload role,
+        // which routes to the dedicated `summarization-v1` tier (PR #2690)
+        // rather than the generic `reasoning-v1` fallback.
         assert_eq!(model, "summarization-v1");
     }
 
@@ -226,6 +226,9 @@ mod tests {
         let mut cfg = Config::default();
         cfg.memory_tree.cloud_llm_model = Some("custom-summary-model".into());
         let (_provider, model) = build_chat_runtime(&cfg).unwrap();
+        // Setting memory_tree.cloud_llm_model overrides the cloud-memory
+        // model path; the routing falls back to the platform default
+        // (`reasoning-v1`) rather than the `summarization-v1` tier.
         assert_eq!(model, "reasoning-v1");
     }