VLMs priced a $2.43 necklace from $19 to $104 across three outfits
BrianneLee011 · hn · 2026-07-28
A Show HN project tested how vision-language models priced a $2.43 necklace across three outfits.
The models produced wildly different estimates, ranging from $19 to $104, making the demo a neat illustration of how VLMs can be sensitive to presentation and context rather than just the object itself.
More from Multimodal
- How Claude Code is being used to orchestrate a six-stage AI video studio — EXM7777 · 2026-07-28
- Google Gemini Omni was used to generate a cinematic alien image and video prompt — michaelrabone · 2026-07-28
- An essay warns AI-generated photos and videos are pushing us into an “anti-reality” age — churchkey · 2026-07-28
- Report: Top Image Editing Models on Hugging Face Easily Generate Nonconsensual Deepfakes — Justgototheeffinmoon · 2026-07-28
- Tom and Jerry in AI live action is so convincing the cartoon looks fake — eyishazyer · 2026-07-28
- AI video workflow locks Seedream 5 Lite references across five shots — LudovicCreator · 2026-07-28