Liquid AI releases LFM3 multimodal model family
What happened
On October 28, 2026, Hacker News' front page reported that Liquid AI released the LFM3 multimodal model family for specialized intelligence. The first batch has six 2B-class models sharing one backbone across text, vision, audio, omni-modal and decision tasks. The new backbone combines short convolutions, sliding-window attention and global attention in a three-part hybrid architecture, pretrained on 22T tokens. The text model offers a 256K-token context window. The report discloses only these six models, their architecture, training scale and text context setup.
Written by AI from the coverage · updated 3 hours ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageIntroducing LFM3: Every Modality for Specialized Intelligence
Liquid AI 发布 LFM3，首批包含 6 款 2B 级模型，以共享骨干覆盖文本、视觉、音频、全模态和决策任务。新骨干采用短卷积、滑动窗口注意力与全局注意力的三混合架构，经过 22T tokens 预训练，文本模型提供 256K tokens 上下文窗口。
Heat over time
Heat now 9·Comparable peak 10（Oct 28 03:00）·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.