SenseNova-U1

5 atoms · 跨 2 天 · 首见 2026-04-29 · 最近 2026-05-24

三色: 🟦 fact 5 · 🟥 take 0 · stance ▲1/▼0/◆0

来源: X 5

时态: fresh:5

标签: 好信源:4 · 好数字:1

事实 (fact) (5)

• 2026-05-24 · matches top understanding-only VLMs in text, vision-language perception, reasoning, agents, and spatial intelligence
  It matches top understanding-only VLMs in text, vision-language perception, reasoning, agents, and spatial intelligence, while excelling in any-to-image synthesis, complex infographics, and interleaved generation.
  tags: 好信源 · → daily
• 2026-05-24 · early results show strong performance in vision-language-action and world model tasks
  Early results also show strong performance in vision-language-action and world model tasks.
  tags: 好信源 · → daily
• 2026-04-29 · 开源
  SenseTime open-sourced SenseNova-U1, a multimodal image generation model built on NEO-Unify!
  tags: 好信源 · → daily
• 2026-04-29 · 模型架构
  This architecture drops the visual encoder and VAE entirely. It generates images natively as one system that can handle understanding, reasoning, and generation processes.
  tags: 好信源 · → daily
• 2026-04-29 · 成本
  Cost: ~1/10 of closed-source alternatives. Pull the weights, host yourself, no per-call pricing.
  tags: 好数字 · → daily
  value: qty=1/10 of closed-source alternatives

时间轴 (近 20)

• 2026-05-24 · fact · matches top understanding-only VLMs in text, vision-language · → daily
• 2026-05-24 · fact · early results show strong performance in vision-language-act · → daily
• 2026-04-29 · fact · 开源 · → daily
• 2026-04-29 · fact · 模型架构 · → daily
• 2026-04-29 · fact · 成本 · → daily

← 实体目录 · 系统日志