header-langage
简体中文
繁體中文
English
Tiếng Việt
한국어
日本語
ภาษาไทย
Türkçe
Scan to Download the APP
Partnered With:

Xiaomi MiMo leads China's large model rankings, smart money bets early and reaps 4x profits.

According to PolyBeats monitoring, Xiaomi's MiMo-V2.6 series, open-sourced on September 22, currently ranks first among all Chinese models on the LMArena text leaderboard. As of now, the probability on Polymarket that "Xiaomi has the best Chinese AI model in October" is 82.5%, up about 22 percentage points over the past week; Moonshot AI is at approximately 12%, and Alibaba at approximately 3.6%.

The resolution rules for this market are to check the overall leaderboard of the arena.ai text arena, using the version with style control turned off and filtered by "Models." Among models primarily belonging to Chinese companies, the market resolves to the company whose model ranks highest.

An account with a win rate exceeding 80% in the model ranking section once bought in when the probability of Xiaomi's model being the best was only 16%, and currently has an unrealized gain of 409% and has not yet taken profit.

0xa2770 bought in $82.4. This account has an 80% win rate across 45 markets traded in the AI ranking section, with PnL +$7.1k and ROI +47.7%. The amount invested in this trade was 0.25 times its average investment in this section.

According to Xiaomi's official website, MiMo-V2.6-Pro is a full-modal mixture-of-experts model with approximately 1.02 trillion total parameters and 42 billion activated parameters, open-sourced under the MIT license; the same series also includes MiMo-V2.6-Flash with 310 billion total parameters and 15 billion activated parameters. Xiaomi took the previous-generation MiMo-V2.5 series offline on October 21.

举报 Correction/Report
RelatedRecommended
Price wars are the final siege tactic of scale monopolists.
A person must put themselves into what size of a machine in order to not waste their intelligence.
Correction/Report
Submit
Add Library
Visible to myself only
Public
Save
Choose Library
Add Library
Cancel
Finish