Issues-translate-bot
commented
1 week ago Bot detected the issue body's language is not English, translate it automatically.
Title: [Feature Request] More flexible visual model discrimination
🥰 Description of requirements
The current project uses fixed keywords and excludes keywords for visual model identification (isVisionModel). In addition, each model manufacturer does not adopt a consistent naming scheme, resulting in lag in model visual identification and frequent modifications, such as the latest gemini-exp- 1114 also supports visual capabilities, but the current visual discrimination cannot be directly adapted, and there is an urgent need to optimize a more flexible visual model discrimination method.
🧐 Solution
Possible solutions:
- Allows adding vision capabilities to the specified model through environment variables, such as: VisionModel=model_1,model_2,model_3
- A model that allows front-end web page configuration and background parsing to support visual capabilities
- ...
📝 Supplementary information
No response
🥰 需求描述
当前项目采用固定的关键词、排除关键词的方案进行视觉模型判别（isVisionModel），加上各模型厂商并没有采取一致的命名方案，导致模型视觉判别滞后和频繁修改，如最新的 gemini-exp-1114 也支持视觉能力了，但是当前的视觉判别不能直接适配，急需优化更灵活的视觉模型判别方法
🧐 解决方案
可能的解决方案：
📝 补充信息
No response