Remalt already has a strong BYOK foundation with OpenRouter, WaveSpeed, Gemini, Anthropic and other providers. I would love to see the provider layer expanded further rather than adding individual AI models one by one.
Please consider adding native BYOK support for:
⦿ fal.ai: Strong production API ecosystem for image, video, audio and multimodal generation
⦿ Replicate, broad catalogue of image, video and open-source models
⦿ Hugging Face Inference Providers - only if this has not already been added
More importantly, it would be valuable if Remalt could dynamically discover and expose compatible models available through connected providers instead of requiring every new model to be manually added by Remalt.
For example, WaveSpeed already provides access to many leading image and video models. If a model is already available through a connected provider, Remalt should ideally expose it automatically rather than requiring another dedicated integration.
The same architecture could support fal.ai, Replicate and Hugging Face as their catalogues evolve.
Ideally the system would:
⦿ Automatically sync newly available provider models
⦿ Clearly show which provider powers each model
⦿ Avoid duplicate model entries across providers
⦿ Let users choose the preferred provider when the same model exists on several APIs
⦿ Show estimated cost/credits before generation
⦿ Support provider-specific image/video parameters rather than only generic prompts
⦿ Fall back to another compatible provider when enabled
This would make Remalt much more future-proof and allow new models to become available much faster without waiting for individual Remalt releases
