Does this support the vision/image inputs for the Qwen models?

[zweily]

I just noticed this project and have this question since the LM studio doesn't even support the models with image/vision capabilities like Qwen3.5/3.6 models to enable the speculative decoding option.

So just wondering if this project supports the image inputs?

Thanks.

[davide221]

Not yet. Luce dflash is text-only right now: target + draft are both Qwen3.5/3.6 text. Any contribution in that direction is welcome! 🚀