diff --git a/backend/ai_services/admin.py b/backend/ai_services/admin.py
index 3da40ef..421a97b 100644
--- a/backend/ai_services/admin.py
+++ b/backend/ai_services/admin.py
@@ -22,8 +22,8 @@ class TranscriptionTaskAdmin(UnfoldModelAdmin):
 
 @admin.register(AIEvaluationTemplate)
 class AIEvaluationTemplateAdmin(UnfoldModelAdmin):
-    list_display = ['name', 'model_selection', 'score_dimension', 'is_active', 'created_at']
-    list_filter = ['is_active', 'model_selection', 'created_at']
+    list_display = ['name', 'model_selection', 'score_dimension', 'is_default', 'is_active', 'created_at']
+    list_filter = ['is_active', 'is_default', 'model_selection', 'created_at']
     search_fields = ['name', 'prompt']
 
 @admin.register(AIEvaluation)
diff --git a/backend/ai_services/migrations/0008_add_is_default_to_template.py b/backend/ai_services/migrations/0008_add_is_default_to_template.py
new file mode 100644
index 0000000..9c0a799
--- /dev/null
+++ b/backend/ai_services/migrations/0008_add_is_default_to_template.py
@@ -0,0 +1,18 @@
+# Generated by Django 6.0.1 on 2026-03-17 15:55
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('ai_services', '0007_aievaluationtemplate_score_dimension'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='aievaluationtemplate',
+            name='is_default',
+            field=models.BooleanField(default=False, help_text='默认模板会评价所有比赛，非默认模板且未关联评分维度时不会自动评价', verbose_name='是否为默认模板'),
+        ),
+    ]
diff --git a/backend/ai_services/models.py b/backend/ai_services/models.py
index 442d913..2e174f4 100644
--- a/backend/ai_services/models.py
+++ b/backend/ai_services/models.py
@@ -73,6 +73,11 @@ class AIEvaluationTemplate(models.Model):
         verbose_name=_('关联评分维度'),
         help_text=_('如果同步到比赛评分，优先使用此维度。未填写则默认使用"AI Rating"或包含"AI"的维度')
     )
+    is_default = models.BooleanField(
+        verbose_name=_('是否为默认模板'),
+        default=False,
+        help_text=_('默认模板会评价所有比赛，非默认模板且未关联评分维度时不会自动评价')
+    )
     is_active = models.BooleanField(verbose_name=_('是否启用'), default=True, help_text=_('启用后，新的转写任务完成后将自动使用此模板进行评估'))
     
     created_at = models.DateTimeField(verbose_name=_('创建时间'), auto_now_add=True)
diff --git a/backend/ai_services/services.py b/backend/ai_services/services.py
index 367360b..cd64249 100644
--- a/backend/ai_services/services.py
+++ b/backend/ai_services/services.py
@@ -331,17 +331,23 @@ class AliyunTingwuService:
     def trigger_ai_evaluations(self, task_id):
         """
         根据启用的模板自动触发 AI 评估
+        
+        逻辑：
+        1. 如果模板关联了评分维度(s score_dimension)，只对关联了相同维度的比赛进行评估
+        2. 如果模板未关联评分维度：
+           - 如果是默认模板(is_default=True)，评价所有比赛
+           - 否则不进行自动评价
         """
         try:
             # 在线程中重新获取 task 对象，并预加载 project，避免懒加载导致的线程数据库连接问题
             from .models import TranscriptionTask
-            task = TranscriptionTask.objects.select_related('project').get(id=task_id)
+            task = TranscriptionTask.objects.select_related('project', 'project__competition').get(id=task_id)
         except Exception as e:
             # 兼容处理：如果 task_id 其实是 task 对象（虽然我们上面改了，但防止其他地方调用传错）
             if hasattr(task_id, 'id'):
                 try:
                     from .models import TranscriptionTask
-                    task = TranscriptionTask.objects.select_related('project').get(id=task_id.id)
+                    task = TranscriptionTask.objects.select_related('project', 'project__competition').get(id=task_id.id)
                 except:
                     task = task_id
             else:
@@ -361,6 +367,41 @@ class AliyunTingwuService:
             if AIEvaluation.objects.filter(task=task, template=template).exists():
                 logger.info(f"Evaluation for task {task.id} and template {template.name} already exists.")
                 continue
+            
+            # 获取任务关联的比赛
+            task_competition = None
+            if task.project and task.project.competition:
+                task_competition = task.project.competition
+            
+            # 判断是否应该对此任务进行评估
+            should_evaluate = False
+            
+            if template.score_dimension:
+                # 模板关联了评分维度，只对关联了相同维度的比赛进行评估
+                if task_competition:
+                    # 获取该比赛下所有关联了相同评分维度的比赛ID列表
+                    from competition.models import ScoreDimension
+                    related_competition_ids = ScoreDimension.objects.filter(
+                        id=template.score_dimension.id
+                    ).values_list('competition_id', flat=True)
+                    
+                    if task_competition.id in related_competition_ids:
+                        should_evaluate = True
+                        logger.info(f"Template '{template.name}' is linked to score_dimension, task's competition matches.")
+                    else:
+                        logger.info(f"Template '{template.name}' is linked to score_dimension, but task's competition does not match. Skipping.")
+                else:
+                    logger.info(f"Task {task.id} has no associated competition. Skipping template '{template.name}'.")
+            else:
+                # 模板未关联评分维度，只有默认模板才评价所有比赛
+                if template.is_default:
+                    should_evaluate = True
+                    logger.info(f"Template '{template.name}' is default template, evaluating all competitions.")
+                else:
+                    logger.info(f"Template '{template.name}' is not linked to score_dimension and is not default. Skipping.")
+            
+            if not should_evaluate:
+                continue
 
             # 创建评估记录
             evaluation = AIEvaluation.objects.create(