Update AI model defaults and tiers

app/routers/ai.py

@@ -11,6 +11,7 @@ from fastapi import APIRouter, Depends, HTTPException, status
 from sqlalchemy import and_, select
 from sqlalchemy.ext.asyncio import AsyncSession
 
+from app.core.config import get_settings
 from app.database import get_db
 from app.models.item import Item
 from app.schemas.ai import (
@@ -21,6 +22,7 @@ from app.schemas.ai import (
     AIStatsResponse,
 )
 from app.services.ai_generation import (
+    SUPPORTED_MODELS,
     generate_question,
     get_ai_stats,
     save_ai_question,
@@ -28,6 +30,7 @@ from app.services.ai_generation import (
 )
 
 logger = logging.getLogger(__name__)
+settings = get_settings()
 
 router = APIRouter(prefix="/admin/ai", tags=["admin", "ai-generation"])
 
@@ -63,14 +66,15 @@ async def generate_preview(
     
     - **basis_item_id**: ID of the sedang-level question to base generation on
     - **target_level**: Target difficulty (mudah/sulit)
-    - **ai_model**: OpenRouter model to use (default: qwen/qwen-2.5-coder-32b-instruct)
+    - **ai_model**: OpenRouter model to use (default: qwen/qwen2.5-32b-instruct)
     """
     # Validate AI model
     if not validate_ai_model(request.ai_model):
+        supported = ", ".join(SUPPORTED_MODELS.keys())
         raise HTTPException(
             status_code=status.HTTP_400_BAD_REQUEST,
             detail=f"Unsupported AI model: {request.ai_model}. "
-            f"Supported models: qwen/qwen-2.5-coder-32b-instruct, meta-llama/llama-3.3-70b-instruct",
+            f"Supported models: {supported}",
         )
 
     # Fetch basis item
@@ -279,14 +283,19 @@ async def list_models() -> dict:
     return {
         "models": [
             {
-                "id": "qwen/qwen-2.5-coder-32b-instruct",
-                "name": "Qwen 2.5 Coder 32B",
-                "description": "Fast and efficient model for question generation",
+                "id": settings.OPENROUTER_MODEL_CHEAP,
+                "name": "Mistral Small 4",
+                "description": "Cheap and fast option for routine variant generation",
             },
             {
-                "id": "meta-llama/llama-3.3-70b-instruct",
+                "id": settings.OPENROUTER_MODEL_QWEN,
+                "name": "Qwen 2.5 32B Instruct",
+                "description": "Balanced default for structured soal generation",
+            },
+            {
+                "id": settings.OPENROUTER_MODEL_LLAMA,
                 "name": "Llama 3.3 70B",
-                "description": "High-quality model with better reasoning",
+                "description": "Premium fallback when you want better quality over cost",
             },
         ]
     }