dots.llm1
MIT-licensed 142B MoE (14B active) trained on 11.2T non-synthetic tokens.
About
The Xiaohongshu/RedNote lab's first open model is a 142B-parameter MoE that activates only 14B per token, released under the MIT license in June 2025. The unusual selling point is transparency: no synthetic data in pretraining, and intermediate checkpoints published every trillion tokens so researchers can study the training dynamics.
RedNote reports performance comparable to Qwen2.5-72B. At 142B total parameters it is not a casual download — you will want serious GPU memory or a quantized build.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- Performance comparable to Qwen2.5-72B (rednote-reported)
- Pretrained on 11.2T high-quality non-synthetic tokens
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: MIT license per press coverage; official HF org rednote-hilab; Released 2025-06-06; no synthetic data in pretraining; checkpoints every 1T tokens; Performance claims are rednote-reported. License: MIT per press coverage of the release; verify on the HF model card..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · MIT license per press coverage; official HF org rednote-hilab
- · Released 2025-06-06; no synthetic data in pretraining; checkpoints every 1T tokens
- · Performance claims are rednote-reported