Rednote's Dots Studio Releases First Open-Source LLM dots3-note: 280B MoE for Long-Horizon Agents
BanghuaZ · x · 2026-08-14
Dots studio lab, affiliated with Rednote (Xiaohongshu), has announced its first language model, dots3-note preview, focusing on long-horizon agentic tasks in real-world environments.
Key Features & Highlights:
- Architecture: A 280B parameter Mixture-of-Experts (MoE) model with 16B active parameters. It supports a 512K context window and offers multimodal understanding across text, vision, and audio.
- New RL Approach: Introduces TEMPO, a novel reinforcement learning method for long-horizon agent training utilizing self-critiquing and test-time-scaled value estimation.
- Agentic Capabilities: Designed to reason, explore unfamiliar environments, update memory over time, and combine multimodal perception with coding and tool use to solve complex tasks.
- Open Source: Model weights are available on Hugging Face, alongside two new open benchmarks for real-life agents: VibeSearchBench and VibeLifeBench.
The studio claims the model is competitive with much larger models across reasoning, agentic, and multimodal tasks.
Related event: Xiaohongshu Open-Sources dots3-note 280B Multimodal Model(14 posts)→
More from Models
- From botching 9.9 vs 9.11 to tackling the hardest math problems in two years — Yuchenj_UW · 2026-10-07
- OpenAI claims 372 unsolved problems cracked, averaging about 3 hours each — i_dg23 · 2026-10-07
- Benchmark: OpenAI Decisions API costs 2x more, 5-10% worse than Jev — xeophon · 2026-10-07
- Cagliostro V3.5 135M Dethrones SmolLM2-135M on Open SLM Leaderboard at 27.49 — Megneous · 2026-10-07
- "If you still think LLMs suck at math, it's a skill issue" — hot take gains traction — alejandroll10 · 2026-10-07
- Sauers_: Opus 5.5 shows fewer strong technical opinions than Fable 5.1 and 5 — Sauers_ · 2026-10-07