While there are AI_IMAGE_PROVIDER and AI_AUDIO_PROVIDER environment variables, the model falls back to AI_TEXT_MODEL, which is not always capable of outputting image or audio.
I suggest AI_IMAGE_MODEL and AI_AUDIO_MODEL are added and used in the respected Ai/Agents files.
This will enable using Ollama or OpenRouter ( pending #220 ) for all AI features in self-hosted installations.
While there are
AI_IMAGE_PROVIDERandAI_AUDIO_PROVIDERenvironment variables, the model falls back toAI_TEXT_MODEL, which is not always capable of outputting image or audio.I suggest
AI_IMAGE_MODELandAI_AUDIO_MODELare added and used in the respectedAi/Agentsfiles.This will enable using Ollama or OpenRouter ( pending #220 ) for all AI features in self-hosted installations.