llama-swap: pin qwen3.5-9b + qwen3.6-35-a3b as a coexistence group

Adds a new `pinned` group with swap: false (models coexist in VRAM),
exclusive: false (group shares with other groups), persistent: true
(never unload). Each member also gets ttl: 0 so the per-model
idle-timeout can't drop them either — belt + suspenders.

Pair is currently qwen3.5-9b (~6 GB Q4) + qwen3.6-35-a3b (~29 GB Q6).
Plus the 128K KV caches, roughly 50-60 GB VRAM resident. Appropriate
for an A6000/H100-class card; verify fit after deploy.

Committed as a canonical change; push + restart still needed on
ana-ml2.

stacks/llama-swap/config.yaml

@@ -24,6 +24,9 @@
 #               Swapped abliterated entry from IIEleven11 Heretic Q8_0 to
 #               mradermacher abliterated i1-Q6_K (already cached).
 #               Bumped both Qwen 3.6 entries ctx-size 32768 → 131072.
+#   2026-04-22: New `pinned` group — qwen3.5-9b + qwen3.6-35-a3b coexist
+#               in VRAM with persistent=true and ttl=0. Means the two
+#               can be called concurrently and never idle-unload.
 # ============================================================================
 
 # Default 1200 seconds (20 min) to wait for model to be available to load.
@@ -97,7 +100,7 @@ models:
   "qwen3.5-9b":
     name: "Qwen 3.5 9B UD-Q4_K_XL"
     description: "Dense 9B model. Lightweight general-purpose chat and reasoning."
-    ttl: 600
+    ttl: 0   # pinned — member of the `pinned` group, never unloads
     cmd: |
       /app/llama-server
       --context-shift
@@ -120,7 +123,7 @@ models:
   "qwen3.6-35-a3b":
     name: "Qwen 3.6 35B-A3B UD-Q6_K_XL"
     description: "Qwen3.6 MoE, 3B active of 35B. Stock unsloth Q6_K_XL (~29GB). Thinking on by default."
-    ttl: 600
+    ttl: 0   # pinned — member of the `pinned` group, never unloads
     cmd: |
       /app/llama-server
       --context-shift
@@ -487,3 +490,15 @@ groups:
       - "embeddinggemma-300M"
       - "qwen3-embedding-0.6B"
       - "qwen3-reranker-0.6B"
+
+  # Pinned general-purpose chat models. Coexist in VRAM, never unload.
+  # Members also have ttl: 0 individually so idle-timeout can't drop them.
+  # VRAM budget check: qwen3.5-9b (~6 GB at Q4) + qwen3.6-35-a3b (~29 GB
+  # at Q6_K_XL) + 128K KV cache for each ≈ 50-60 GB. Size for your GPU.
+  "pinned":
+    swap: false
+    exclusive: false
+    persistent: true
+    members:
+      - "qwen3.5-9b"
+      - "qwen3.6-35-a3b"