feat(avatar): add private vision chat support

backend/app/api/endpoints/ai_models.py

@@ -37,6 +37,8 @@ async def create_model(req: AIModelCreateRequest, db=Depends(get_db)):
         api_base_url=req.api_base_url,
         api_key_enc=encrypt(req.api_key) if req.api_key else None,
         model_version=req.model_version,
+        vision_model_version=req.vision_model_version,
+        ocr_model_version=req.ocr_model_version,
         temperature=req.temperature,
         max_tokens=req.max_tokens,
         timeout_seconds=req.timeout_seconds,
@@ -100,6 +102,8 @@ async def get_digital_avatar_runtime_model(
         "api_base_url": model.api_base_url or "https://api.openai.com/v1",
         "api_key": decrypt(model.api_key_enc) if model.api_key_enc else "",
         "model": model.model_version or model.model_name,
+        "vision_model": model.vision_model_version or "qwen3.6-flash",
+        "ocr_model": model.ocr_model_version or "qwen-vl-ocr",
         "temperature": model.temperature,
         "max_tokens": model.max_tokens,
         "timeout_seconds": model.timeout_seconds,
@@ -129,6 +133,8 @@ def _format_model(m: AIModelConfig) -> dict:
         "usage_scope": m.usage_scope,
         "api_base_url": m.api_base_url, "has_api_key": bool(m.api_key_enc),
         "model_version": m.model_version, "temperature": m.temperature,
+        "vision_model_version": m.vision_model_version,
+        "ocr_model_version": m.ocr_model_version,
         "max_tokens": m.max_tokens, "timeout_seconds": m.timeout_seconds,
         "is_default": m.is_default, "is_enabled": m.is_enabled,
         "created_at": m.created_at.isoformat(),