Rename 'vision' to 'image_to_text' for consistency

coderai

@@ -46,7 +46,7 @@ from dataclasses import dataclass, field
 class ModelCapabilities:
     """Represents what a model can do."""
     text_generation: bool = False  # LLM/chat completion
-    vision: bool = False  # Image understanding
+    image_to_text: bool = False  # Image understanding (captioning, VQA)
     image_generation: bool = False  # Text-to-image (Stable Diffusion)
     speech_to_text: bool = False  # Audio transcription
     text_to_speech: bool = False  # Speech synthesis
@@ -55,8 +55,8 @@ class ModelCapabilities:
         caps = []
         if self.text_generation:
             caps.append("text")
-        if self.vision:
-            caps.append("vision")
+        if self.image_to_text:
+            caps.append("image-to-text")
         if self.image_generation:
             caps.append("image")
         if self.speech_to_text:
@@ -83,9 +83,9 @@ def detect_model_capabilities(model_name: str) -> ModelCapabilities:
         caps.image_generation = True
         return caps  # Usually SD models are dedicated
     
-    # Check for vision models
+    # Check for vision models (image-to-text)
     if any(x in name_lower for x in ['vision', 'vl-', '-vl', 'llava', 'qwen2-vl', 'qwen-vl', 'phi-4-mini', 'pixtral', 'clip']):
-        caps.vision = True
+        caps.image_to_text = True
         caps.text_generation = True  # Vision models are also LLMs
         return caps