diff --git a/stacks/qwen35-vl/compose.yaml b/stacks/qwen35-vl/compose.yaml
index 5377998..3cbbb59 100644
--- a/stacks/qwen35-vl/compose.yaml
+++ b/stacks/qwen35-vl/compose.yaml
@@ -54,6 +54,9 @@ services:
       - ${QWEN_MAX_MODEL_LEN}
       - --dtype
       - auto
+      # Prefix caching pinned ON (the nightly defaults it OFF). Free win for the
+      # text-chat path; marginal for vision (each image is a distinct prefix).
+      - --enable-prefix-caching
     deploy:
       resources:
         reservations:
diff --git a/stacks/vllm/compose.yaml b/stacks/vllm/compose.yaml
index afe3d7e..186ee4b 100644
--- a/stacks/vllm/compose.yaml
+++ b/stacks/vllm/compose.yaml
@@ -243,6 +243,11 @@ services:
       # FP8 KV cache — halves KV memory; near-lossless on Ada (cc 8.9).
       - --kv-cache-dtype
       - ${GRANITE_KV_CACHE_DTYPE}
+      # Prefix caching pinned EXPLICIT (vLLM v1 defaults it on, but pin so a
+      # version flip can't silently disable it). Benched 2026-06-13: ~6.5x faster
+      # TTFT (45ms vs 292ms) on a shared ~4.5k-token summarizer template; soft/
+      # evictable KV, neutral when prefixes don't repeat — pure win for granite.
+      - --enable-prefix-caching
     deploy:
       resources:
         reservations: