llama-swap: add Qwen3.6 35B-A3B stock + swap abliterated, 128K ctx

- New: qwen3.6-35-a3b using -hf unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q6_K_XL
  (~29 GB weights, thinking on, KB Qwen sampler defaults).
- Swap: qwen3.6-35-a3b-abliterated from IIEleven11 Heretic Q8_0 to
  mradermacher/Qwen3.6-35B-A3B-abliterated-i1-GGUF:i1-Q6_K (already in
  the /hfcache on ana-ml2; ~27 GB).
- Context: bumped both Qwen 3.6 entries 32768 -> 131072.

stacks/llama-swap/config.yaml

@@ -20,6 +20,10 @@
 #   2026-04-17: Added Qwen3.6-35B-A3B Abliterated Heretic Q8_0 via -hf syntax.
 #               Requires HF_HOME=/hfcache in compose (see docker-compose.yml).
 #               New convention: use -hf repo[:quant] instead of --model /path.
+#   2026-04-20: Added stock unsloth Qwen3.6-35B-A3B at Q6_K_XL via -hf syntax.
+#               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.
 # ============================================================================
 
 # Default 1200 seconds (20 min) to wait for model to be available to load.
@@ -113,18 +117,40 @@ models:
   # Qwen 3.6 — uses -hf syntax, reads from HF_HOME=/hfcache (host pre-download)
   # --------------------------------------------------------------------------
 
-  "qwen3.6-35-a3b-abliterated":
-    name: "Qwen 3.6 35B-A3B Abliterated Heretic Q8_0"
-    description: "Qwen3.6 MoE, 3B active. Abliterated/heretic variant of BF16 quantized to Q8_0. ~38GB."
+  "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
     cmd: |
       /app/llama-server
       --context-shift
       --jinja
-      -hf IIEleven11/Qwen3.6-35B-A3B-Abliterated-Heretic-BF16-Q8_0-GGUF
+      -hf unsloth/Qwen3.6-35B-A3B-GGUF:UD-Q6_K_XL
       --port ${PORT}
       --n-gpu-layers 999
-      --ctx-size 32768
+      --ctx-size 131072
+      --flash-attn on
+      --temp 1.0
+      --top-p 0.95
+      --top-k 20
+      --min-p 0.00
+      --presence-penalty 1.5
+      --repeat-penalty 1.0
+      --reasoning on
+      --reasoning-format deepseek
+
+  "qwen3.6-35-a3b-abliterated":
+    name: "Qwen 3.6 35B-A3B Abliterated i1-Q6_K"
+    description: "Qwen3.6 MoE, 3B active of 35B. mradermacher abliterated imatrix Q6_K (~27GB)."
+    ttl: 600
+    cmd: |
+      /app/llama-server
+      --context-shift
+      --jinja
+      -hf mradermacher/Qwen3.6-35B-A3B-abliterated-i1-GGUF:i1-Q6_K
+      --port ${PORT}
+      --n-gpu-layers 999
+      --ctx-size 131072
       --flash-attn on
       --temp 1.0
       --top-p 0.95