Xiaomi mở mã nguồn hơn 7.000 môi trường MiMo RL và code huấn luyện trên Hugging Face

Nguồn chính

1. SOURCE@artificialanlys · 20:11 21 thg 9 · “it retains the same attractive pricing at $0.435 per 1M input tokens (with a 99% cache-hit discount) and $0.87 per 1M output tokens” · x.com
2. SOURCE@victormustar · 20:18 21 thg 9 · “Pro scores 46.32 on the Artificial Analysis Intelligence Index, the highest of any open model to date” · x.com
3. SOURCE@eliebakouch · 2:19 26 thg 9 · “xiaomi MiMo oss RL environments + training code” · x.com
4. SUPPORT@nrehiew_ · 1:48 21 thg 9 · “The pro model went from 58.41 to 72.57 on DeepSWE” · x.com
5. SUPPORT@_luofuli · 22:27 21 thg 9 · “we have released a Qwen model distilled from MiMo RL trajectories as a stronger starting point for RL, along with 7K diverse environments and a complete RL training framework” · x.com
6. SUPPORT@teortaxestex · 3:37 22 thg 9 · “The team estimates a 10× productivity gain, shortening the R&D cycle from one month to 2–3 days” · x.com
7. SUPPORT@teortaxestex · 21:29 21 thg 9 · “This RL run had cost peanuts, 5 days of a ≈10K GPU cluster, it didn't even target CritPt explicitly” · x.com
8. SUPPORT@kimmonismus · 22:14 21 thg 9 · “Better than Grok 4.7, much cheaper and holy moly is china back. That is the real surprise!” · x.com