Launch
lawrencewzen/pi-vlm-proxy: 让非多模态模型(DeepSeek 等)把图片识别委托给任意 OpenAI 兼容的多模态模型。零厂商硬编码,主模型支持视觉时自动隐藏自己。
github.com
GitHub · 29 stars · posted by lawrencewzen
A pi extension adding a describe_image tool so non-multimodal models can delegate image recognition to any OpenAI-compatible VLM.
Launch
github.com
GitHub · 29 stars · posted by lawrencewzen
First spotted Aug 3, 2026 on GitHub