Unverified50% confidenceFactExact time

Current multimodal large models including GPT-4o and Claude 3.5 Sonnet can 'see' images but encode them into visual tokens processed at a semantic level rather than perceiving pixel-level details

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/2/2026

First Seen

Valid until: 9/30/2026

Sources

Cursor Fails at UI Design Reproduction: The Real Capability Boundaries of AI Coding

bilibili明天玩点什么6/3/2026

Related Claims

UnverifiedGPT-4V、Gemini Vision、Claude等多模态大模型的视觉理解建立在视觉编码器（通常为Vision Transformer）与大规模语言模型深度融合的架构之上76% similar VerifiedGPT-4 Vision Preview是在纯文本的GPT-4基础上外挂了一个视觉编码器模块，图片需先经过视觉模块处理再传入语言模型76% similar VerifiedGPT-4o结合视觉编码器的多模态架构能在统一的语义空间中对齐不同类型的信息74% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/53566

API

curl https://kongchang.com/api/v1/knowledge/claims/53566

MCP

get_claim(id=53566)