Any plans about paper-reader Agent with multimodal LLM?

[zoubaihan]

Your project is very interesting and useful, but as we all know, many papers contain flowcharts and illustrations. Would you consider using a multimodal large model to read papers in the future?

[InMatrix]

@zoubaihan Thank you for your feedback. The project currently uses Gemini 2.5 Flash (upgrading to 3.0 soon) to summarize PDF files. The model is indeed multimodal, so I'd expect it to incorporate information from charts and figures in its output.

Related, I'm considering having the model identify the most important figure from the paper and then include it in the paper summary. It's probably doable but will likely require a paid API key.