diff --git a/stacks/llama-swap/conf/config.yaml b/stacks/llama-swap/conf/config.yaml
index 09b035d..1ac775a 100644
--- a/stacks/llama-swap/conf/config.yaml
+++ b/stacks/llama-swap/conf/config.yaml
@@ -320,7 +320,7 @@ models:
 
   "artemis-31b-v1i":
     name: "Artemis 31B v1i Q6_K"
-    description: "BeaverAI Artemis-31B-v1i — Gemma 4 31B dense fine-tune, Q6_K (~28.6GB). 256K context."
+    description: "BeaverAI Artemis-31B-v1i — Gemma 4 31B dense fine-tune, Q6_K (~28.6GB). 256K context native; served at 128K."
     ttl: 600
     cmd: |
       /app/llama-server
@@ -328,13 +328,15 @@ models:
       -hf BeaverAI/Artemis-31B-v1i-GGUF:Q6_K
       --port ${PORT}
       --n-gpu-layers 999
-      --ctx-size 32768
+      --ctx-size 131072
       --flash-attn on
       --temp 1.0
       --top-p 0.95
       --top-k 64
       --repeat-penalty 1.0
       --chat-template-kwargs '{"enable_thinking":true}'
+      --reasoning on
+      --reasoning-format deepseek
 
   # ==========================================================================
   # GLM MODELS