MiMo

Headlines · 6

1. 2d
   Xiaomi HyperOS 4 begins second stable rollout to more devices

   Xiaomi announced that the stable version of HyperOS 4 is beginning a second rollout to additional phones, tablets, and televisions. The first batch includes Xiaomi 17 Max, Xiaomi 17T Pro, Xiaomi 17T, several REDMI phone and tablet models, and multiple Xiaomi TV series. Xiaomi says the update integrates Super XiaoAI 2.0 and its self-developed MiMo large model, alongside improvements in cross-device coordination and system fluidity.

   IT之家 AI · 🔥 3
2. 5d
   Why AI Product Retention May Outweigh Leaderboard Wins

   In an authorized TMTPost analysis, Zhixun Think Tank argues that model quality can win trials but does not necessarily create lasting user loyalty. It recommends comparing candidates on real task delivery, rework time, full cost, workflow integration, and continuity, rather than relying only on leaderboard positions.

   钛媒体 · 🔥 0
3. Oct 22
   Study traces the origins of “wait” tokens in reasoning models

   A paper under review by Kou and eight coauthors argues that reinforcement learning does not necessarily teach reasoning models to write “wait” before reconsidering a step. Across analyzed open models, the marker often appeared during mid-training or reasoning-trace fine-tuning, while reinforcement learning substantially improved MATH-500 accuracy without introducing it in models whose bases lacked the word. The authors also report that some reinforcement-learning recipes multiplied markers already present in the base, so the effect depends on the recipe. They further warn that marker frequency alone does not demonstrate reinforcement-learning-induced reasoning and that LLM-as-a-judge revision metrics may reverse human-labeled trends.

   Hugging Face · Papers · 🔥 0
4. Oct 15
   Xiaomi HyperOS 4 starts first stable rollout with Super Xiao Ai 2.0 and MiMo AI

   IT之家 reported on October 15 that Xiaomi began the first stable rollout of HyperOS 4, covering the Xiaomi 17 series (including Xiaomi 17 Ultra, 17 Pro Max, 17 Pro and 17), the REDMI K100 Pro Max, K100 Pro, K90 Pro Max and K90, and the Xiaomi Pad 8 series. The report says the release upgrades AI capabilities, cross-device collaboration and baseline smoothness, integrating Super Xiao Ai 2.0 with Xiaomi's in-house MiMo large model and adding features such as an expert mode and the Inspiration Ball interaction; Xiaomi's own claim is that the system is smoother and delivers a more fluid experience. Later batches are listed as extending by October 31 to the Xiaomi 17 Max, 17T series and REDMI K90 Max, reaching the Xiaomi 15 and REDMI K80 series by mid-November, the Xiaomi 14/13 series plus more tablets, TVs and wearables within December 2026, and some TVs and bands in January 2027.

   IT之家 AI · 🔥 0
5. Oct 10
   QuotaPanel v0.2.0 adds Xiaomi MiMo quota source

   QuotaPanel is an open-source desktop panel that consolidates subscription and API balances from multiple AI providers into a single view. Version 0.2.0 moves quota information into the system menu bar and adds Xiaomi's MiMo Token Plan as a quota source. According to OSChina, the update also aims to make it easier to trace the root cause of unusual issues.

   开源中国 · 🔥 0
6. Oct 6
   Post-90s leaders take over China's LLM race: Qwen, MiMo, Tencent, ByteDance

   In September 2026 Alibaba confirmed Liu Dayiheng (b. 1993) as head of the ATH group's TokenFoundry Qwen large-language-model project ahead of its Yunqi conference, and Chen Yusen (b. 1992) appeared as an Alibaba group vice president and CEO of the Qwen Office unit. The same day Xiaomi released the MiMo-V2.6 series and promoted Luo Fuli (b. 1995) to level 22, the top of Xiaomi's job ladder; her MiMo-V2.6-Pro scored 46 on Artificial Analysis's composite intelligence index, which the article says makes it the top open-weight model on that list. It also describes Zhou Chang running ByteDance's Seedream, Seedance and world-model work and Yao Shunyu heading both AI Infra and the LLM department at Tencent since December 2025, arguing the job handover is complete while capability, commercialization and cost tests are only beginning.

   36氪 人工智能 · 🔥 0

Experience and discussion from the community

Share my MiMo experienceAsk about MiMo