以下部分引用 AI 总结，现阶段 AI 仍然有幻觉。请以内容的原文为准。
数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15
NVDA Cosmos 3
10 atoms · 跨 1 天 · 首见 2026-06-01 · 最近 2026-06-01
三色: 🟦 fact 7 · 🟥 take 3 · stance ▲7/▼0/◆0
来源: X 10
时态: fresh:10
标签: 好信源:4 · 好数字:3 · 好观点:2 · 好思考:1
叙事 (narrative) (1)
- 2026-06-01 · is foundation for building intelligence for autonomous machines
Cosmos 3 is not just a model release. It is the foundation for building intelligence for autonomous machines.
tags: 好观点 · → daily
事实 (fact) (9)
- 2026-06-01 · uses dual-tower Mixture-of-Transformers architecture
Cosmos 3 uses a dual-tower Mixture-of-Transformers architecture.
tags: 好信源 · → daily
- 2026-06-01 · is fully open-source
Cosmos 3 is fully open.
tags: 好信源 · → daily
- 2026-06-01 · can understand and generate across language · images · video · audio · action sequences
It can understand and generate across: language · images · video · audio · action sequences
tags: 好信源 · → daily
- 2026-06-01 · available under Linux Foundation’s OpenMDW 1.1 License
Everything is available under the Linux Foundation’s OpenMDW 1.1 License.
tags: 好信源 · → daily
- 2026-06-01 · ranked #1 open model on Artificial Analysis Image-to-Video leaderboard
#1 open model on Artificial Analysis Image-to-Video leaderboard
tags: 好数字 · → daily
- 2026-06-01 · ranked #1 open model on Artificial Analysis Text-to-Image leaderboard
#1 open model on Artificial Analysis Text-to-Image leaderboard
tags: 好数字 · → daily
- 2026-06-01 · ranked #1 open model on Physics-IQ
#1 open model on Physics-IQ, which measures accuracy on physical laws
tags: 好数字 · → daily
- 2026-06-01 · training action domains together creates positive transfer
training action domains together creates positive transfer.
tags: 好思考 · → daily
- 2026-06-01 · is first omni-model for physical AI
Cosmos 3 is the first omni-model for physical AI.
tags: 好观点 · → daily
时间轴 (近 20)
- 2026-06-01 ·
fact · ranked #1 open model on Artificial Analysis Image-to-Video l · → daily
- 2026-06-01 ·
fact · ranked #1 open model on Artificial Analysis Text-to-Image le · → daily
- 2026-06-01 ·
fact · ranked #1 open model on Physics-IQ · → daily
- 2026-06-01 ·
fact · uses dual-tower Mixture-of-Transformers architecture · → daily
- 2026-06-01 ·
fact · is fully open-source · → daily
- 2026-06-01 ·
fact · is first omni-model for physical AI · → daily
- 2026-06-01 ·
fact · can understand and generate across language · images · video · → daily
- 2026-06-01 ·
fact · training action domains together creates positive transfer · → daily
- 2026-06-01 ·
fact · available under Linux Foundation’s OpenMDW 1.1 License · → daily
- 2026-06-01 ·
narrative · is foundation for building intelligence for autonomous machi · → daily
← 实体目录 · 系统日志