Xiaohongshu Open-Sources a 280B Multimodal Model
💡A 280B open-weight MoE with 512K context could reshape multimodal content AI development.
⚡ 30-Second TL;DR
What Changed
dots3-note preview uses a Mixture-of-Experts architecture with 280B total and 16B activated parameters.
Why It Matters
The release shows a major content platform treating foundation models as core infrastructure rather than a standalone feature. Its permissive license, long context, and multimodal capabilities could make it relevant for developers building content, recommendation, and agent workflows.
What To Do Next
Clone the dots3-note preview repositories and run a small multimodal evaluation against your current model on long-context content tasks.
Key Points
- •dots3-note preview uses a Mixture-of-Experts architecture with 280B total and 16B activated parameters.
- •It supports a 512K context window and multimodal understanding across text, vision, and voice.
- •The model is released under Apache 2.0 on Hugging Face and GitHub.
- •Huawei Ascend support was adapted on day zero, broadening potential deployment options.
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: Pandaily ↗
This is a summary, not the original. Read the source, or get the weekly briefing.
Weekly AI briefing
One email a week. Unsubscribe anytime.