diff --git a/studio/backend/core/training/worker.py b/studio/backend/core/training/worker.py
index 8eee8293e1..1496c8255b 100644
--- a/studio/backend/core/training/worker.py
+++ b/studio/backend/core/training/worker.py
@@ -198,9 +198,23 @@ def run_training_process(
             if trainer.should_stop:
                 event_queue.put({"type": "complete", "output_dir": None, "ts": time.time()})
             else:
+                error_msg = trainer.training_progress.error or "Failed to load model"
+                # Hint about trust_remote_code if YAML says this model needs it
+                if not config.get("trust_remote_code", False):
+                    try:
+                        from utils.models.model_config import load_model_defaults
+                        model_defaults = load_model_defaults(model_name)
+                        yaml_trust = model_defaults.get("training", {}).get("trust_remote_code", False)
+                        if yaml_trust:
+                            error_msg = (
+                                f"Model '{model_name}' requires trust_remote_code to be enabled. "
+                                f"Please enable 'Trust remote code' in Chat Settings and try again."
+                            )
+                    except Exception:
+                        pass
                 event_queue.put({
                     "type": "error",
-                    "error": trainer.training_progress.error or "Failed to load model",
+                    "error": error_msg,
                     "stack": "", "ts": time.time(),
                 })
             return
diff --git a/studio/backend/routes/inference.py b/studio/backend/routes/inference.py
index 67c858f032..263700a68c 100644
--- a/studio/backend/routes/inference.py
+++ b/studio/backend/routes/inference.py
@@ -222,27 +222,28 @@ async def load_model(
                 except Exception as e:
                     logger.warning(f"Could not read adapter_config.json: {e}")
 
-        # Resolve trust_remote_code: use True if either the request or YAML config says so.
-        # This ensures models like Nemotron that require it always get it, even if the
-        # frontend toggle hasn't been set yet.
-        trust_remote_code = request.trust_remote_code
-        if not trust_remote_code:
-            model_defaults = load_model_defaults(config.identifier)
-            yaml_trust = model_defaults.get("inference", {}).get("trust_remote_code", False)
-            if yaml_trust:
-                logger.info(f"YAML config sets trust_remote_code=True for {config.identifier}")
-                trust_remote_code = True
-
         # Load the model
         success = backend.load_model(
             config=config,
             max_seq_length=request.max_seq_length,
             load_in_4bit=load_in_4bit,
             hf_token=request.hf_token,
-            trust_remote_code=trust_remote_code,
+            trust_remote_code=request.trust_remote_code,
         )
 
         if not success:
+            # Check if YAML says this model needs trust_remote_code
+            if not request.trust_remote_code:
+                model_defaults = load_model_defaults(config.identifier)
+                yaml_trust = model_defaults.get("inference", {}).get("trust_remote_code", False)
+                if yaml_trust:
+                    raise HTTPException(
+                        status_code=400,
+                        detail=(
+                            f"Model '{config.display_name}' requires trust_remote_code to be enabled. "
+                            f"Please enable 'Trust remote code' in Chat Settings and try again."
+                        ),
+                    )
             raise HTTPException(
                 status_code=500,
                 detail=f"Failed to load model: {config.display_name}"
diff --git a/studio/backend/routes/training.py b/studio/backend/routes/training.py
index 151f8d8542..b6e6d7e111 100644
--- a/studio/backend/routes/training.py
+++ b/studio/backend/routes/training.py
@@ -19,14 +19,12 @@ if str(backend_path) not in sys.path:
 # Import backend functions
 try:
     from core.training import get_training_backend
-    from utils.models.model_config import load_model_defaults
 except ImportError:
     # Fallback: try to import from parent directory
     parent_backend = backend_path.parent / "backend"
     if str(parent_backend) not in sys.path:
         sys.path.insert(0, str(parent_backend))
     from core.training import get_training_backend
-    from utils.models.model_config import load_model_defaults
 
 # Auth
 from auth.authentication import get_current_subject
@@ -196,14 +194,6 @@ async def start_training(
             "trust_remote_code": request.trust_remote_code,
         }
 
-        # Resolve trust_remote_code: use True if either the request or YAML config says so.
-        if not training_kwargs["trust_remote_code"]:
-            model_defaults = load_model_defaults(request.model_name)
-            yaml_trust = model_defaults.get("training", {}).get("trust_remote_code", False)
-            if yaml_trust:
-                logger.info(f"YAML config sets trust_remote_code=True for {request.model_name}")
-                training_kwargs["trust_remote_code"] = True
-
         # Free GPU memory: shut down any running inference/export subprocesses
         # before training starts (they'd compete for VRAM otherwise)
         try: