SenseTime co-founder: encoder-free native VLMs key step toward unified multimodal AI
liuziwei7 · x · 2026-10-19
SenseTime co-founder liuziwei7 commented that native, encoder-free VLMs are an important step toward more unified multimodal intelligence. He noted SenseTime is pioneering the same paradigm with its NEO and SenseNova-U model series, sharing links to both, and said he's excited to see the field move forward together.
More from Models
- Retrained "bee model" drifts into dreamy, detached trace-like outputs — Sauers_ · 2026-10-19
- One prompt in a normal ChatGPT chat yields an 87-second chalkboard explainer video — mallow610 · 2026-10-19
- Claude Opus normal chats burn code quota, user says ChatGPT sub is far more generous — mallow610 · 2026-10-19
- A ~56KB meditative HTML page one-shot with Gemini using Antigravity — Saboo_Shubham_ · 2026-10-19
- OpenAI models top Artificial Analysis token burn at 197k output tokens per task — teortaxesTex · 2026-10-19
- Redditor begs OpenAI for a work-only ChatGPT plan, says Anthropic's allowances beat it for agents — bofors14 · 2026-10-19