chore(ai): replace deprecated llama models with qwen3.6-27b and gpt-oss-120b

Author: yash-pouranik

.github/scripts/generate-changelog-entry.js

@@ -48,7 +48,7 @@ Rules:
 7. If no PRs were merged (or if all were ignored), output: <Update label="${TODAY}" description="Week of ${WEEK_START}–${TODAY}" tags={[]}>No significant changes this week.</Update>`;
 
 const payload = JSON.stringify({
-  model: "llama-3.3-70b-versatile",
+  model: "qwen/qwen3.6-27b",
   messages: [{ role: "user", content: prompt }],
   max_tokens: 10000,
   temperature: 0.3,

apps/dashboard-api/src/controllers/ai.controller.js

@@ -246,7 +246,7 @@ const collectionCreator = async (req, res, next) => {
             developer_id: req.user._id.toString(),
             plan: effectivePlan,
             encrypted_byok: encryptedByok,
-            model: model || 'llama-3.3-70b-versatile'
+            model: model || 'qwen/qwen3.6-27b'
         });
 
         const allowedTypes = new Set(['String', 'Number', 'Boolean', 'Date', 'Object', 'Array', 'Ref']);

apps/python-service/routers/ai.py

@@ -81,7 +81,7 @@ class CollectionCreatorRequest(BaseModel):
     developer_id: str
     plan: str = "free"
     encrypted_byok: EncryptedByok | None = None
-    model: str = "llama-3.3-70b-versatile"
+    model: str = "qwen/qwen3.6-27b"
 
 class CollectionCreatorResponse(BaseModel):
     type: Literal["clarify", "schema", "complete"]

apps/python-service/services/byok.py

@@ -41,7 +41,7 @@ async def resolve_ai_client(
     encrypted_byok: dict | None,
     feature: str = "query-builder",
     increment: bool = True,
-    model: str = "llama-3.3-70b-versatile",
+    model: str = "qwen/qwen3.6-27b",
 ) -> ChatGroq:
     """Resolve the appropriate AI client based on BYOK and plan.
 

apps/python-service/tests/test_byok.py

@@ -46,7 +46,7 @@ async def test_byok_key_decrypted_and_used(mock_decrypt):
         mock_decrypt.assert_called_once_with(encrypted_byok["groqKey"])
         MockGroq.assert_called_once_with(
             api_key="gsk_test_byok_key_12345",
-            model_name="llama-3.3-70b-versatile",
+            model_name="qwen/qwen3.6-27b",
             temperature=0,
         )
 
@@ -73,7 +73,7 @@ async def test_free_tier_under_limit(mock_redis):
 
         MockGroq.assert_called_once_with(
             api_key=settings.GROQ_API_KEY,
-            model_name="llama-3.3-70b-versatile",
+            model_name="qwen/qwen3.6-27b",
             temperature=0,
         )
 
@@ -100,7 +100,7 @@ async def test_pro_tier_under_limit(mock_redis):
 
         MockGroq.assert_called_with(
             api_key=settings.GROQ_API_KEY,
-            model_name="llama-3.3-70b-versatile",
+            model_name="qwen/qwen3.6-27b",
             temperature=0,
         )
 
@@ -111,7 +111,7 @@ async def test_pro_tier_under_limit(mock_redis):
         client = await resolve_ai_client("dev123", "pro", None)
         MockGroq.assert_called_with(
             api_key=settings.GROQ_API_KEY,
-            model_name="llama-3.3-70b-versatile",
+            model_name="qwen/qwen3.6-27b",
             temperature=0,
         )
 

apps/web-dashboard/src/components/CollectionCreatorAgent.jsx

@@ -13,8 +13,6 @@ const SUGGESTIONS = [
 ];
 
 const MODELS = [
-  'llama-3.3-70b-versatile',
-  'llama-3.1-8b-instant',
   'qwen/qwen3.6-27b',
   'openai/gpt-oss-120b',
   'openai/gpt-oss-20b',