DeepSeek is still not multimodal, and that gap makes little sense, says commentator

dotey · x · 2026-07-24

The post argues that DeepSeek still not supporting multimodal inputs is hard to understand, and notes that the founder’s earlier skepticism toward coding agents has since changed.

It is framed as a reaction to a longer talk by Liang Wenfeng, but the concrete takeaway is a critique of DeepSeek’s current model capability gap on multimodality.

Related event: DeepSeek's Liang Wenfeng Pledges Sole Focus on AGI Over Profits(18 posts)→

Original post →

More from Models

Models channel →