ask: switch to llama3.2:latest, increase timeout to 120s

gemma3:latest produces garbage output on the Vulkan backend (Intel Arc A380).
llama3.2:latest runs correctly at 100% GPU. Timeout bumped to 120s to handle
cold model loads (~22s) without timing out.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

matrixbot/commands.py

@@ -873,7 +873,7 @@ async def cmd_ask(client: AsyncClient, room_id: str, sender: str, args: str):
     await send_text(client, room_id, "Thinking...")
 
     try:
-        timeout = aiohttp.ClientTimeout(total=90)
+        timeout = aiohttp.ClientTimeout(total=120)
         async with aiohttp.ClientSession(timeout=timeout) as session:
             async with session.post(
                 f"{OLLAMA_URL}/api/chat",

matrixbot/config.py

@@ -20,7 +20,7 @@ LOG_LEVEL = os.getenv("LOG_LEVEL", "INFO")
 OLLAMA_URL = os.getenv("OLLAMA_URL", "http://10.10.10.157:11434")
 OLLAMA_MODEL = os.getenv("OLLAMA_MODEL", "llama3.2:latest")
 BALL_MODEL = os.getenv("BALL_MODEL", "sadiq-bd/llama3.2-1b-uncensored:latest")
-ASK_MODEL = os.getenv("ASK_MODEL", "gemma3:latest")
+ASK_MODEL = os.getenv("ASK_MODEL", "llama3.2:latest")
 MINECRAFT_RCON_HOST = os.getenv("MINECRAFT_RCON_HOST", "10.10.10.67")
 MINECRAFT_RCON_PORT = int(os.getenv("MINECRAFT_RCON_PORT", "25575"))
 MINECRAFT_RCON_PASSWORD = os.getenv("MINECRAFT_RCON_PASSWORD", "")